Health and the democratic dividend in Sub-Saharan Africa: are democracies better at managing
the HIV/AIDS epidemic?
Demi Joszefine van de Nes S1839004
Dr. Leila Demarest
Bachelor project 10: practicing democracy in contemporary Africa
Word count: 8165 17/06/2019
Abstract
During the 1990s many African countries experienced political transitions towards democracy. During this same period, many African countries also experienced an outbreak of HIV/AIDS. According to the democratic dividend theory, democracies should provide socio-economic benefits to their citizens and improve their standards of living. Scholars suggest that the legitimacy of African democracies is partly dependent on this dividend. If one were to follow this theory, democracies should handle the HIV/AIDS epidemic better than non-democracies because of their higher level of accountability. In this paper, a cross-national analysis is conducted to investigate how a country’s level of democracy is associated with respondents’ knowledge of HIV/AIDS and the provision of HIV-testing places. Data from the Demographic and Health Surveys (DHS) conducted in Sub-Saharan Africa as well as country-level data is used in a multilevel analysis. The relationships between both democracy and knowledge of HIV and democracy and the provision of HIV-testing places are non-significant. In other words, democracy does not affect HIV/AIDS policies and democracies do not handle the HIV/AIDS epidemic better than non-democracies. This can have implications for how citizens view their legitimacy.
Introduction
It has been one of Africa’s greatest leadership challenges: the fight against the HIV/AIDS epidemic. For many African leaders, the first reaction to the AIDS epidemic was denialism (Kagaayi & Serwadda, 2016). Nowadays, for some leaders, this has not changed. According to former South-African president Mbeki, AIDS is not caused by HIV, but by poverty. His ‘solution’ was fighting poverty instead of providing medicines (Gwaambuka, 2016; Boseley, 2008). His successor, Jacob Zuma, who was accused of rape during his time as vice-president, said that he took a shower to lower the risk of getting HIV (BBC News, 2006; Volkskrant, 2006). Not only South-African leaders have non-scientific ideas about the causes, transmission, and treatment of HIV. In 2004, the Ugandan president Museveni openly announced a ‘war against condoms’. From his point of view, condoms did not protect against HIV, but promoted promiscuity (New Vision, 2004; Schoepf, 2004). In 2014, Museveni signed a bill which criminalizes the transmission of HIV and allows institutions to disclose the HIV-status of infected citizens without their permission (Human Rights Watch, 2014). The former Gambian president Jammeh argued that he could cure AIDS using his alleged mystic powers (Duval Smith, 2007).
These ideas against the scientific consensus concerning AIDS are devastating, certainly in Sub-Saharan Africa, the region that has been hardest hit by the epidemic (UN AIDS, 2018). Having and sharing correct knowledge about the cause, transmission and treatment of the disease is key in fighting the epidemic (World Health Organization, 2019). Denialism, the spread of false information and the implementation of incorrect policies do have a major impact (Boseley, 2008). Yet is there a difference amongst different regimes in handling the HIV/AIDS epidemic?
Sub-Saharan Africa has a great variety of different regime types. While decolonization and democratization started during the 1950s, most African countries became authoritarian during the 1960s. During the 1990s, re-democratization started off (Bratton, Mattes, Gyimah-Boadi, 2005; Bratton & Van der Walle, 1997; Cheeseman, 2015). Not all African countries went through the same process and even less of them were eventually successful in their route to democracy. Setbacks towards authoritarian rule due to military coups, civil wars and ignored election results were no exception and continue to be a cause for concern today. The legitimacy of Africa’s current democracies is partly dependent on the ability to provide a ‘democratic dividend’ to their citizens (Ake, 1993; Cheeseman, 2015; Masaki & Van de Walle, 2014).
According to Bratton, Mattes and Gyimah-Boadi (2005), the democratic dividend concerns the benefits that come with democracy. Cheeseman (2015) stresses that democracies implement more policies which benefit their citizens, due to the incentives the government faces. These beneficial policies combined with more accountability and the transparency of their leadership will result in a more effective and responsible government. The dividend should increase the living conditions and welfare of the citizens in African countries.
The democratic dividend should also involve healthcare. As Sub-Saharan Africa is most severely affected by HIV/AIDS (UN AIDS, 2018), an increase in living conditions and the improvement of healthcare should also involve HIV/AIDS policies. Although new HIV infections are declining in Sub-Saharan Africa, transmission and deaths still occur (Harsch, n.d.; Mail & Guardian, 2011; UN AIDS, 2018). As mentioned, sharing correct information and providing treatment is key in the fight against the epidemic (World Health Organization, 2019). From the perspective of a democratic dividend, one can expect a relationship between democracy and such beneficial HIV/AIDS policies. If this relationship is not found, one could not only question the health policies in Sub-Saharan Africa, but also the legitimacy of new African democracies. Focusing on this, this paper does not only analyze the relationship between democracy and HIV/AIDS policies in Africa but also questions whether current African democracies are able to provide a democratic dividend.
The research question this paper therefore addresses is whether democracies are better at handling the HIV/AIDS epidemic than non-democracies. To investigate the relationship between regime type and HIV/AIDS policies, a cross-national analysis focusing on citizens’ knowledge concerning AIDS and the provision of HIV-testing places is conducted. This empirical analysis differs from previously conducted research because it is solely focused on Sub-Saharan Africa and makes use of micro-level data from the Demographic and Health Surveys (DHS). These data are combined with country-level data, including democracy scores, in multilevel analyses.
Results indicate that there are no significant relationships between democracy and the respondent’s knowledge of HIV/AIDS and democracy and the provision of HIV-testing places. Therefore, one can conclude that African democracies do not handle the HIV/AIDS epidemic better than non-democracies, which can impact how African citizen’s see the legitimacy of their democracy.
The paper proceeds as follows. Section 1 explains the AIDS crisis in Sub-Saharan Africa. Section 2 discusses the related literature and sets out the hypotheses. Section 3 explains the methodology and the data. Section 4 presents the empirical results and finally, section 5 concludes the paper.
1. Case study: The AIDS crisis in Sub Saharan Africa
Acquired Immune Deficiency syndrome (AIDS) is the final phase of the Human Immunodeficiency Virus (HIV). HIV belongs to the family of retroviruses; a type of virus that integrates into the cells of their host (International Partnership for Microbicides, n.d.-a). In the case of HIV, the virus takes over the T cells, which are a part of the immune system. These cells replicate themselves and ‘attack’ the non-infected T cells. The infected cells do not work properly and die off, which will reduce the amount of T cells in the host’s body. Due to this process, it will be hard for the immune system to fight any diseases and infections (Avert, 2018; Avert, 2019-a; Sharp et al., 2001).
HIV gets transmitted through body fluids, such as blood, semen, vaginal fluids and breastmilk (hiv.gov, 2018). After the infection, the virus will develop through three stages. The first stage is the acute HIV infection, which takes place 2 – 4 weeks after the infection and results in flu symptoms. Already in this stage, HIV takes over the T cells and reduces the amount in the body. However, at the end of the first stage, the amount of T cells will be relatively stable (Avert, 2019-a). The second stage is the asymptomatic phase, in which HIV is still present in the body but the carrier is not showing any symptoms. During this stage, the virus does replicate but at very low levels. Nonetheless, the level of T cells is still declining, which does damage the immune system badly. This makes the host vulnerable to diseases and illnesses. If the HIV carrier has an illness, the virus has developed to the third stage: AIDS. The development of the second to the third stage can take years (Avert, 2019-a).
There are several types of HIV, all of which can result in AIDS, if left untreated. Overall, there are two main types: HIV-1 and HIV-2, which both consist of multiple groups. HIV-1 is the most common and most infectious. HIV-2 is rare and less infectious than the former type. The different strains of HIV are the result of the origin of the disease (Sharp & Hahn, 2011).
HIV originates from monkeys, which were carriers of a similar virus to HIV. Chimpanzees were carriers of Simian Immunodeficiency Virus (SIV), which is the source of HIV-1. Humans got infected with SIV due to their hunting and consumption of apes: the ‘bushmeat’ theory (Sharp et al., 2001). The SIV virus adapted to the human body, which was the start of our current epidemic. The different groups within HIV-1 are due to various times of transmission from monkey to humans. The SIV virus kept developing within the monkey's body and with new infectious contact between an infected primate and a non-infected human, a new type of HIV was transmitted. HIV-2 is related to a similar virus found in sooty mangabey apes. The ‘bushmeat theory’ also applies to this type of HIV (Sharp et al., 2001).
Kinshasa is seen as the starting point of the epidemic (Sharp et al., 2001; Sharp & Hahn, 2011). Not only because the first HIV infected blood samples were found there, but also because most of the HIV strains are present in the area. This would suggest that monkeys with various forms of SIV lived in the area and transferred their virus to humans there. However, HIV did not stay in West-Central Africa. Due to transportation, trade and sexual activities, HIV spread silently throughout the continent and the world (Sharp & Hahn, 2011).
Worldwide, HIV and AIDS are still spreading with their epicenter in Sub-Saharan Africa (UN AIDS, 2018). Histories of homosexual activity or drug abuse are rarely reported in Africa. More common modes of transmission in Sub-Saharan Africa are heterosexual transmission, injections, transfusions and perinatal transmission (Buvé, Bishikwabo-Nsarhaza, & Mutangadura, 2002; Quinn, Mann, Curran & Piot, 1986). Due to biological differences and gender inequalities, women are more vulnerable to HIV than men (Avert, 2019-b; International Partnership for Microbicides, n.d.-b). However, there are global differences. In Europe, the female-to-male ratio is 1:3, while the ratio in West and East Africa 1.22:1 is and in Southern Africa 1.33:1 (UN AIDS, 2010; World Health Organization, 2018). According to UN AIDS (2006), 76% of all HIV infected women are living in Sub-Saharan Africa, where they make up for 59% of the infected population.
The high percentage of prevalence in Africa is partly due to the poor promotion of condom use. According to Buvé et al. (2002), in 1997, only 28% of Cameroonian sex workers reported condom use with their last client. In contrast, in Thailand, on the second hardest hit continent, 78% of the sex workers were consistent in their condom use. Besides the lack of
condom use amongst sex workers, premarital sex is also more common in Sub-Saharan Africa than in Asia. More (unsafe) sexual partners enlarge the risk of infection (Buvé et al., 2002). Moreover, the first reaction for many African leaders was the denial of the threat, which resulted in a lack of needed HIV prevention programs (Kagaayi & Serwadda, 2016). Even if the leaders wanted to tackle the problem during the beginning of the epidemic, checking blood transfusions on T cells infected with HIV and treating AIDS patients was too expensive for the African countries (Quinn et al., 1986; Söderlund, Lavis, Broomberg & Mills, 1993; World Health Organization, 2019).
There are not only inter-continental differences, but also within Africa, differences in the spread of HIV/AIDS are found (Figure 1). The transmission has been more rapid in Southern and Eastern Africa, than in Western and Central Africa. This is not only due to sexual behavior patterns but also to social and economic patterns: urbanization, modernization, poverty levels, the decline of social services and the differences in gender power. Furthermore, many African conflicts and wars play a part in the spread (Buvé et al., 2002; Kagaayi & Serwadda, 2016).
Figure 1. HIV prevalence in Sub-Saharan Africa in 2017 (World Bank Indicators. (n.d.))
1,9 1 22,8 0,8 3,7 4 1,3 0,7 3,1 2,8 0,6 0,9 4,2 1,7 1,5 4,8 1,4 0,3 1,2 0,3 12,5 12,1 0,3 2,8 0,4 1,4 0,1 18,8 2,4 0,2 4,5 5,9 11,5 13,3 0,1 27,4 HIV prevalence (%)
Until today, the epidemic has not been defeated (UN AIDS, 2018). Awareness of the disease and treatment are key in the fight. Only by avoiding the transmission, HIV, and eventually AIDS can be defeated (World Health Organization, 2019). Therefore, this thesis focuses on two key factors in avoiding HIV transmission: the respondent’s awareness of AIDS and the provision of HIV-testing places.
2. Theoretical Framework
The basis of the relationship between democracy and healthcare policies can be found in two main theories. These theories argue that democracies have higher government accountability and provide more public goods. The two theories are explained below and applied to the HIV/AIDS epidemic.
The main thought of the first theory is that democracies have higher government accountability than non-democracies. According to Przeworski and Cheibub (1999), regimes are accountable if the citizens can punish their leaders if they do not act in their best interest. If the government satisfies its citizens it stays in office, if it does not, it will get punished. Governments which survival in office depends on the permission of the citizens are accountable governments. Therefore, democracy should enforce a level of accountability. Due to elections, the electorate can easily punish a government for their decisions. While in an autocracy it takes more effort for the citizens to punish their governments.
Based on the argumentation of Przeworksi and Cheibub (1999), one could argue that there is a principal-agent game, where the principal (electorate) can punish the agent (government). This leads to a higher level of accountability in democracies than in non-democracies. If this theory would be applied to the HIV/AIDS epidemic, one could argue that democracies, with a higher level of accountability, would implement more effective HIV/AIDS awareness policies than non-democracies. If democracies do not do this, the government can get penalized by their electorate, while in non-democracies the likelihood of this punishment is lower.
Also, accountability has to do with the provision of public goods. Research has shown that overall, democracies provide more public goods than autocracies (Besley & Kudamatsu, 2006; Deacon, 2009; Rosenzweig, 2015; Wigley & Akkoyunlu-Wigley, 2011). Democracies are
forced by their electoral process and institutions to provide public goods. Lake and Baum (2001) argue that every politician aims utility maximization, but politicians in democracies are bound by their electorate. Non-democratic states are less bound by the people they oversee and therefore they will use their monopoly to earn more rents and provide fewer goods. In democratic states, the opposite will occur. Democracies will use fewer monopoly rents and provide more public goods. The institutional context is different between democracies and non-democracies, which will result in variance considering the provision of public goods.
Similar results are presented by Rosenzweig (2015), who claims that more electoral competition will lead to a higher provision of public goods, regardless of countries’ GDP per capita and the democratic level of elections. According to Bueno de Mesquita, Smith, Siverson, & Morrow (2003), democratic leaders maintain their position by allocating public and private goods by coalition size. A smaller winning coalition needs to produce more public goods to stay in power than a bigger winning coalition. Deacon (2009) stresses that the range of supporters is crucial. In democracies, the government needs to satisfy a large group of the population, wherein autocracies only a smaller, targeted group needs to be appeased. Similar to Deacon, Wigley and Akkoyunlu-Wigley (2011) emphasize the range of supporters which need to be pacified. Therefore, democracies would provide more health-promoting resources than non-democracies. If one were to apply the public goods theory on the HIV/AIDS epidemic, one would expect that democracies will provide more public goods than non-democracies.
Adsera, Boix and Payne (2003) stress that accountability does not only depend on free elections, but also on the level of information that is available for the electorate. A well-informed electorate has more precise knowledge of the adopted and implemented policies, which can help the electorate to punish the leaders in the next elections. If the electorate lacks this information, there is less reason for them to punish their governments (Adsera et al., 2003).
Key in a well-informed electorate is freedom of the press. The research above is mainly focused on developed societies but the theory of accountability could also be applied to less developed societies. Sen (1982, 1999) argues that famines are less common in democracies because the poor and most affected people have the chance to penalize the politicians for their suffering. Sen claims this is mostly due to the free press. According to him, in the case of famine, the free press should, and would, make it a public issue (Sen, 1994). Creating a public issue would enlarge the knowledge of the electorate and thus the accountability of the government.
D'Souza (1994) agrees with Sen. D'Souza concludes that the Chinese Famine 1959-61 was a direct result of the lack of press freedom by withholding information and active censorship. D'Souza also focused on India; a country relatively vulnerable to famines. Surprisingly, India faces relatively few famines. This can be explained by India being a democracy (I) and having free press (II). The political future of India's politicians depends on disasters such as famines, therefore they will take careful decisions to avoid such disasters.
The accountability theory can be applied to the HIV/AIDS epidemic as well. Ruger (2005), for instance, stresses that public information is crucial to fight HIV/AIDS and that the media plays a central role in sharing information. The press can create awareness concerning the AIDS epidemic and the measures taken by the government. By doing so, the government will be forced to take action to prevent electoral loss. Events such as this happened in South Africa, where the Treatment Action Campaign successfully forced the government to provide anti-retroviral drugs (Treatment Action Campaign, n.d.). Applying the accountability theory to the HIV/AIDS epidemic would create the expectation that countries with higher freedom of the press would have more effective policies on HIV/AIDS than countries lacking press freedom. The free press could make the HIV/AIDS epidemic a public issue. According to Sen (1994), the free press would provide information, which would not be available in a country without freedom of the press. This information would inform the government and have an impact on government policies. In other words, the freedom of the press would inform the government on the HIV/AIDS situation in the county and push the government towards policies that create awareness of the virus and provide treatment to infected citizens1.
Wigley and Akkoyunlu-Wigley (2011) also stress the importance of media freedom for the provision of public goods. Free media would result in better informed people who are able to criticize the government’s policies and it would enable the flow of information of the poor and affected areas to the government officials. Therefore, free media should increase the provision of public goods (Wigley and Akkoyunlu-Wigley, 2011)
This theoretical overview leads towards four different hypotheses divided in two sections according to the dependent variables. The respondent’s awareness of HIV/AIDS is the first dependent variable. This variable contains two different hypotheses:
1 Freedom of the press could also lead to more information about HIV/AIDS provided by the media and given to
Hypothesis I: Due to the enlarged level of accountability, citizens in democracies have a higher knowledge of HIV/AIDS than citizens in non-democratic countries.
Hypothesis II: Citizens of countries which have higher freedom of the press have more knowledge of HIV/AIDS than citizens living in countries lacking press freedom.
The second section of hypotheses are according to the dependent variable of provided HIV-testing places:
Hypothesis III: Democratic countries will provide more HIV/AIDS-testing facilities than non-democratic countries.
Hypothesis IV: Countries which have higher freedom of the press will provide more HIV/AIDS-testing facilities than countries lacking press freedom.
The relationship between democracy and health has been researched before but results differ (Franco, Álvarez-Dardet, & Ruiz, 2004; Ross, 2006; Safaei, 2006; Van der Windt & Vandoros, 2017). According to Ross (2006), democracy has little or no effect on child and infant mortality. Van der Windt & Vandoros (2017) argue in their case-study on Congo’s local leaders that democracy indeed has little to no effect on the health of the citizens. Franco et al. (2004) show a positive relationship between democracy and health. By using a cross-national analysis with country-level variables, they argue that democracy has an independent positive relationship with health, as measured by life expectancy, infant mortality and maternal mortality, even after adjustment for economic factors. A similar relationship has been found by Safaei (2006), who conducts a multi country analysis with country-level variables. Important to take note of is that all this research has been done with country-level variables only. Safaei (2006) stresses the same point and argues that further research on a cross-national level with population health surveys and measures on physical morbidity is necessary. This thesis will fill this gap and combine data from health surveys with country-specific measures, focusing specifically on the HIV/AIDS epidemic.
The relationship between democracy and the AIDS epidemic has also already received researchers’ attention (Justesen, 2012; Marsaudon & Thuillez, in press), but most research is focused on one-country cases or has worldwide coverage. According to Marsaudon and
Thuillez (in press), who conducted a longitudinal study with survey data on Kenya, a person living under democracy is less likely to be infected with HIV than a person not living under democracy. Justesen (2012) makes a similar conclusion while conducting a cross-national study including countries from Latin America, Africa and South-East Asia. Justesen (2012) also argues that democracies with proportional election systems increase access to treatment of HIV/AIDS. Whereas Marsaudon and Thuillez use two rounds of the DHS data sets, Justesen works again with country-level variables. This thesis will combine these two data sources to conduct a cross-national study including survey and country-level data.
This thesis focusses specifically on Saharan Africa, motivated by two reasons; 1) Sub-Saharan Africa is the region hardest hit by HIV/AIDS. According to the UN AIDS (2018), approximately 70% of all HIV infected people are living in Sub-Sahara Africa; 2) Sub-Saharan Africa has a great variation in democratic and non-democratic regimes (Freedom House Index, 2018). The high level of infected inhabitants and the variation of regimes makes Sub-Saharan Africa an excellent region to research the relationship between democracy and healthcare further.
3. Methodology and data
To test the hypotheses set out in the previous section, a quantitative analysis is conducted which covers thirty-two different African countries to investigate the effect of countries’ democracy ratings and press freedom on citizens’ knowledge of HIV/AIDS and their access to healthcare facilities. The analysis will make use of multilevel logistic regression models.
The data used for the analysis is collected at the individual level as well as on the country level. The individual-level variables are derived from the Demographic and Health Surveys (DHS), which are nationally representative multi-country surveys which collect data on a wide range of indicators concerning, amongst others, health (DHS, n.d.-b). DHS has different sorts of surveys, but for this analysis, the standard DHS survey is used. The standard DHS survey involves data on HIV/AIDS-related knowledge, attitudes and behavior and HIV prevalence (DHS, n.d.-b). DHS is representative for the national level, the urban-rural level and the regional level. Sampling occurs in multiple stages, variations in sampling probabilities are corrected via weights (DHS, 2018b; DHS, n.d.-a).
The surveys used in this analysis were conducted between 2008 and 2017, which means that it involves three different rounds of the DHS survey: round 7, round 6 and round 5. For each of the thirty-two countries, the most recent standard DHS survey is used2. These surveys are
divided into a separate dataset for men (ages 15-64) and women (ages 15-64). The countries for which only one of these datasets was available (Sudan), for which the most recent dataset has a large time gap compared to the other countries (Central African Republic, 1994), or for which the needed information is not included (Niger, Tanzania, Sao Tomé and Principe, South Africa) are left out of the analysis. All other datasets – country-specific men and women – are merged together in one Sub-Saharan DHS dataset with a total of 675,834 respondents34.
The first hypothesis concerns the relationship between democracy and the knowledge of HIV/AIDS. The dependent variable is called ‘Ever heard of AIDS?’. Respondents are asked if they have ever heard of the disease AIDS. The answer options to the question are ‘yes’ (1) and ‘no’ (0), which makes the variable dichotomous. Therefore, a logistic regression analysis is chosen. Table 1 shows the mean, standard error, minimum and maximum value of the dependent variable of all countries together. The total mean of ‘Ever heard of AIDS’ is 0.9531 or 95.31%, which indicates that in most countries a large majority of respondents have already heard of AIDS. Out of all six democratic countries, three (Ghana, Lesotho and Namibia) score above the total mean5. The other three countries (Benin, Mali and Sierra Leone) score under
the total mean. The minimum value belongs to undemocratic Chad (0.77, SE 0.004). Amongst the two countries with freedom of the press, only Ghana (0.97, SE 0.157) scores above the total mean. The maximum value of 1.00 belongs to Eswatini, Kenya, Rwanda and Uganda, all undemocratic countries. All of these countries do have the maximum possible score, if rounded on two decimals. Only less than 0.5 percent of the respondents out of these countries never heard of AIDS. This makes the high score of Eswatini (1.00, SE 0.001) remarkable. Eswatini is not only rated as a non-democratic country but also has the highest percentage of HIV prevalence (25.5%) (World Bank Indicators, n.d.). Despite, or perhaps due to, Eswatini’s high level of HIV prevalence and their status as non-democratic country, most respondents do know the disease AIDS.
2 All datasets are free downloadable after registration at DHS:
https://dhsprogram.com/data/new-user-registration.cfm
3 The dataset contains more women (462,070) than men (213,764). Therefore, the weights are corrected,
resulting in men and women have an equal influence (DHS, 2018b).
4 For more information on the dataset, See appendix 1.
The independent variable is the Freedom House Index, one year prior to the conducted survey (FHI_1), which is derived from the Freedom House website (2018). The Freedom House Index combines political rights and civil rights. The average rate measures the democratic level of countries on a scale of 1 (most democratic) to 7 (least democratic). Within the rating, three different levels of democracy can be distinguished: Free (0 – 2.5), Partly Free (3.0 – 5.0) and Not Free (5.5 – 7.0). The ratings are done by a team of expert advisors and external analyst using 25 methodological questions (Freedom House, 2019). Freedom House Index contains the 32 Sub-Saharan countries in our analysis and has a mean of 4.38 (SE 0.264), which is considered to be partially free (table 1). The minimum value belongs to Ghana (1.5, Free) and the maximum value belongs to Burundi, Chad and Cote D’Ivoire (6.5, Not free).
The second hypothesis concerns the relationship between freedom of the press and the knowledge of HIV/AIDS. The dependent variable is again ‘Ever heard of AIDS’. The independent variable is the Freedom of the Press Index, one year prior to the conducted survey (Press_1). This variable is also derived from Freedom House (2017b) and has a 0 (best) to 100 (worst) scale, based on 23 methodological questions. The process of rating the countries is done by a team of analysts and experts and the rating is divided into three different categories of freedom: Free (0 - 30), Partly Free (31 - 60) and Not Free (61 - 100) (Freedom House, 2017a). In our analysis, the mean of the Freedom of the Press Index is 58.38 (SE 2.873), which is considered to be partially free. The minimum value belongs to Mali (24, Free) and the maximum value belongs to Gambia and the Democratic Republic of Congo (83, Not free). The correlation coefficient of Freedom House and the freedom of the press is high, but not perfect (r = 0.857 p < 0.01)6. This means that free democracies do not always have a free press, which
is why it is necessary to regard them as separate measures.
The third hypothesis focuses on the relationship between democracy and the provision of services. The used dependent variable is ‘Do you know a place for a HIV test?’, which is again a dichotomous variable with answer options ‘yes’ (1) and ‘no’ (0). This question is only asked to respondents who have already heard of AIDS. In each DHS survey, the respondent is asked whether they know a place where they can get an HIV test. These facilities could be private places or public places. The. Table 1 shows the mean, standard error, minimum and maximum value of the dependent variable of each country. The total mean is 0.7578 (SE 0.0324) or
75.78%. Of all the democratic countries, three countries (Ghana, Lesotho and Namibia) score above this value, meaning that respondents in these countries know more HIV-testing facilities than the average of all countries together. Overall, if a democratic country scores above average concerning HIV knowledge, it will score above average concerning testing facilities too, and vice versa. Notable is the minimum value, which belongs to Mali (0.39; SE 0.488), a democratic country.
The variable ‘Do you know a place for a HIV test?’ does not measure the absolute number of provided testing places, but whether the respondent knows any facilities. Therefore, the variable ‘Knowledge of HIV’ will be included in this analysis too. This variable is based on nine different variables which are focused on the transmission of HIV and include the following questions: ‘Can you get HIV by mosquito bites (1), sharing food (2) or witchcraft (3)’, ‘Can HIV be transmitted during pregnancy (4), delivery (5) or breastfeeding (6)’ and ‘Can you reduce the chance of getting HIV by not having sexual intercourse at all (7), using a condom every time you have sex (8) or having just one uninfected sexual partner (9)’. The answers to these questions are recoded from yes (1)/no (0) to right answer (1)/wrong answer (0). The variable ‘Knowledge of HIV’ takes an average of these answers, because not all nine questions are asked in each survey. The more right answers a respondent gives, the higher the average score is and the more knowledge the respondent has on HIV/AIDS transmission. By including this variable, the dependent variable can be controlled for a respondent’s overall knowledge of HIV. In this way, ‘Do you know a place for a HIV test?’ measures the actual provision of HIV-testing facilities better.
To test the third hypothesis, the independent variable is again Freedom House Index, one year prior to the survey (FHI_1), which is discussed above. The fourth hypothesis focuses on the relationship between freedom of the press and the provision of public services. The used dependent variable is ‘Do you know a place for a HIV test?’ and the independent variable is Freedom of the Press Index. Again, the variable ‘Knowledge of HIV’ will be included.
All analyses include a number of control variables. These are divided into country-specific variables and individual variables. There are four country-specific control variables. First, the HIV prevalence of the total population measured one year prior to the survey. This refers to the percentage of people (ageing 15 to 49) infected with HIV in the country (World Bank Indicators, n.d.). Countries with a high percentage of HIV prevalence could have taken more measures against HIV/AIDS than countries with a lower HIV prevalence. This could result in more facilities for HIV-testing and more policies on creating awareness on AIDS. Besides, in countries with a low HIV prevalence rate, the need for awareness and demand for HIV-testing places is probably lower. Therefore, it is important to include HIV prevalence as a control variable. Second, the population of each country, one year prior to the survey calculated in natural log. This is the total population, including all residents regardless of their legal status. The value is a midyear value (World Bank Indicators, n.d.). This variable is included because it could be argued that informing citizens or providing testing facilities to a bigger population is harder than to a smaller population. Third, the land area measured one year prior to the survey calculated in natural log. This variable is measured in square kilometers and excludes all area under inland water bodies and national claims on continental territory (World Bank Indicators, n.d.). Land area is included for the same reason as population. It could be harder for countries to provide testing places or awareness programs to their people if they are spread out over a larger area. Fourth, GNI per capita based on the based on purchasing power parity (PPP, current international $), measured one year prior to the survey (World Bank Indicators, n.d.) and calculated in natural log. One could argue that a higher GNI per capita (PPP), could point to a richer society, which has more policies on the awareness of HIV/AIDS and more money for testing facilities.
Table 1. Descriptive statistics of dependent and independent variables
Mean SE Min Max N
‘Ever heard of AIDS?’ 0.9531 0.0094 0.77 1.00 32
‘Do you know a place for a HIV test?’
0.7578 0.0324 0.39 0.99 32
Freedom House Index 4.3781 0.2644 1.5 6.5 32
Freedom of the Press Index 58.3750 2.8729 24 83 32
There are eight individual control variables, starting with gender: women are more vulnerable to HIV/AIDS than men and in Sub Saharan Africa, women are still being treated unequally compared to men (UN AIDS, 2018; Mutume, n.d.). Second, the risk group of HIV is young adults between 15 to 26, thus age is included as a control variable. Third, the place of living is important. Therefore, there is a divide between urban and rural places. One could argue that testing places and government policies on HIV awareness are less present in rural places because these are poorer, more difficult to reach, and generally have a lower number of inhabitants. Fourth, literacy: illiterate respondents have less access to written mediums of the press, such as newspapers, internet and leaflets. Fifth, marriage could have an influence on HIV/AIDS awareness because married people are less likely to have a wide range of different sexual partners, which makes them less prone to HIV/AIDS. Sixth, the number of sexual partners is included because the more sexual partners one has, the more chance on HIV contamination. Seventh, one could argue that the more education one has, the more knowledge one has about diseases such as HIV/AIDS. Therefore, the control variable of education in years is included in the analysis. The final control variable is religion. For religious reasons sickness could be considered as a punishment of God. Therefore, taking precautions would not have an effect on the treatment of HIV (Zou et al., 2009). This variable made the distinction between not religious (0) and religious (1).
4. Results of the cross-national analysis
The results of the multi-level regression analysis on the dependent variable ‘Ever heard of
AIDS?’ are summarized in table 2. In all three models, the two main independent variables do
not have a significant influence on the respondent’s awareness of AIDS; therefore, we cannot reject the null hypothesis of no effect.
According to model 1, the respondent’s knowledge of AIDS is not affected by an increase in the Freedom House Index. This confirms the null hypothesis of no effect. Therefore, this result does not support hypothesis I: Due to the enlarged level of accountability, citizens in
democracies have a higher knowledge of HIV/AIDS than citizens in non-democratic countries.
Most of the country-level variables do not have a significant effect on the dependent variable. An increase of one unit in the population, land area or GNI per capita, will not affect the
respondent’s awareness of AIDS. These relations are not in line with the expectations. However, the HIV prevalence rate has a positive significant effect on the awareness of the respondent; if the percentage of HIV prevalence increases with one unit, odds of ‘Ever heard
of AIDS’ are 1.063 times higher. The intra-class correlation is 0.2485, which means that 24.85%
of the remaining variance can be explained by differences between countries.
On the individual level, all variables have a significant result. If there is an increase in years concerning someone’s age or education, the odds of ‘Ever heard of AIDS?’ are, respectively, 1.032 and 1.203 times higher. The odds of married people are higher too than those of who are not married and people who can read have higher odds concerning their knowledge of AIDS than those who cannot. The place of residence and gender, both have a negative significant effect on ‘Ever heard of AIDS?’; the odds of women are 0.724 times higher compared to those of men and the odds of respondents in the rural areas are 0.387 times higher than those in the urban areas. Lastly, the number of sexual partners has a positive effect; if the respondent has one more sexual partner (excl. their spouse) during the last 12 months, the odds of ‘Ever heard
of AIDS’ are 1.866 times higher. Noticeable, is the variable religion; the odds of a religious
respondent are 1.603 times higher than a non-religious respondent. This does not comply with the theory of Zou et al. (2009). All other control variables have the expected effects.
The second model shows similar results than the first model. Again, the relationship between the main independent variable, freedom of the press, and the dependent variable is not significant. Therefore, the null hypothesis cannot be rejected, and again the results do not confirm hypothesis II: Citizens of countries which have higher freedom of the press have more
knowledge of HIV/AIDS than citizens living in countries lacking press freedom. The intra-class
correlation is 0.2497, which means that 24.97% of the remaining variance can be explained by differences between countries.
The country-level variables have a similar effect as in model one. Again, population, land area and GNI per capita, have no significant negative relationship with the respondent’s knowledge of HIV and the HIV prevalence rate has a significant positive relationship. On individual level, the effects are similar too.
The third model shows the regression if Freedom House Index and Freedom of the Press are both included in the analysis. Again, the relationships are not significant in this model.
Therefore, a relationship between democracy, freedom of press and knowledge concerning HIV is not supported. The country-level variables and the individual level variables have similar effects as in the previous two model. The intra-class correlation is 0.2557, which means that 25.57% of the remaining variance can be explained by differences between countries.
Tab le 2. M u lti -l eve l r egr es si on an al ys is on “Eve r h ear d of A ID S ” M ode l 1 M ode l 2 M ode l 3 b Int erc ept 3. 195 (4 .211) 3. 155 (4 .427) 4. 032 (4 .497) F re edom H ous e (O ne ye ar pri or t o s urve y) 1. 068 (0 .096) 1. 135 (0 .193) F re edom P re ss (O ne ye ar pri or t o s urve y) 1. 00 4 (0 .010) 0. 993 (0 .021) Count ry -l ev el v ar iabl es P opul at ion (N atu ra l L og ) 1. 309 (0 .249) 1. 309 (0 .246) 1. 316 (0 .244) L and A re a (N at ura l L og ) 0. 700 (0 .195) 0. 704 (0 .189) 0. 692 (0 .189) G N I pe r c api ta , P P P (N at ura l L og ) 1. 006 (0 .279) 1. 009 (0 .287) 0. 996 (0 .290) H IV pre va le nc e (%) 1. 063* (0 .025) 1. 062* (0 .026) 1. 064* (0 .025) Indi vi dual -l ev el v ar iabl es A ge 1. 032*** (0 .002) 1. 032*** (0 .002) 1. 032*** (0 .002) E duc at ion i n ye ars 1. 203*** (0 .019) 1. 203*** (0 .019) 1. 203*** (0 .019) Re li gi on (Re f. N ot re li gi ous ) 1. 603** (0 .147) 1. 603** (0 .147) 1. 603** (0 .147)
G ende r (Re f. M al e) 0. 724** (0 .103) 0. 724** (0 .103) 0. 724** (0 .103) M ari ta l S ta tus (Re f. N ot m arri ed) 2. 045*** (0 .063) 2. 045*** (0 .062) 2. 045*** (0 .063) L it era cy (Re f. Ca n’t re ad) 1. 949*** 1. 949*** 1. 949*** (0 .071) (0 .071) (0 .071) P la ce of re si de nc e (Re f. U rba n) 0. 387*** (0 .116) 0. 387*** (0 .116) 0. 387*** (0 .116) N um be r of s ex ua l pa rt ne rs (e xc l. s pous e) 1. 866** (0 .204) 1. 866** (0 .204) 1. 866** (0 .204) Re si dua l V ari anc e 3. 29 0 a 3. 29 0 3. 29 0 Int erc ept V ari anc e 1. 088 1. 095 1. 130 IC C (%) 24 .85% 24 .97% 25 .57% No te : M ul ti-Le ve l Logi st ic Re gr es si on odds r at io coef fici ent s wi th s ta nd ar d er ro rs b et we en b ra ck et s. D ep en de nt v ar ia ble is ‘ E ve r h ea rd o f A ID S’ w ith th e r efe re nc e cat egor y: N o. *** p < 0 .001, ** p < 0 .01, * p < 0 .05 a " # ( $ Sni jde rs & B os ke r, 1999 ) b T he mo de l i s te st ed o n mu lti co lli ne ar ity , S ee a pp en di x 4
Table 3 summarizes the multi-level logistic regression analysis on the respondent’s knowledge of HIV-testing places. The Freedom House Index and Freedom of the Press both do not have a significant relationship with the dependent variable. If the Freedom House Index increases with one unit, the odds of ‘Know a place to get HIV test’ are not affected (model 4). This means that there is no relationship between democracy and the provision of HIV testing facilities. Therefore, the null hypothesis cannot be rejected and hypothesis III: Democratic countries will
provide more HIV/AIDS-testing facilities than non-democratic countries cannot be confirmed.
The intra-class correlation is 0.2837, which means that 28.37% of the remaining variance can be explained by differences between countries.
On the country-level, only the HIV prevalence rate has a significant effect; if the HIV prevalence rate goes up with one unit, the odds of ‘Know a place to get HIV test’ are 1.120 times higher. Population, land area and GNI per capita, do not have a significant effect on the dependent variable.
On the individual level, almost all variables have a significant effect. If there is an increase in years in someone’s age or education, the odds of ‘Know a place to get HIV test’ will be higher with respectively 1.020 and 1.164 times. Moreover, the odds of religious people knowing a place to get a HIV test are 1.433 times higher than the odds of non-religious people. A positive significant relationship can also be found between the marital status of the respondent and the dependent variable. Also, the odds of respondents living in rural areas are 0.584 times higher than the odds of respondents living in urban areas. An increase of sexual partners or the knowledge of HIV transmission does have a positive effect on the respondent’s knowledge of a HIV testing place too. Notably, there is a non-significant relationship of literacy and gender on the respondent’s knowledge of a HIV testing place. This differs from the results in table 2 and is not in line with the expected effects.
The fifth model shows a non-significant relationship between the freedom of the press and ‘Know a place to get HIV test’. If the freedom of the press increases with one unit, the odds of the dependent variable are not affected. In other words, the respondent’s knowledge concerning places to test on HIV is not influenced by the freedom of the press. This relationship does not confirm the hypothesis: Countries which have higher freedom of the press will provide more
HIV/AIDS-testing facilities than countries lacking press freedom. This does not comply with
the provision of public goods. On the contrary, due to the lack of a significant relationship between the freedom of the press and the provision of facilities for HIV testing, the null hypothesis cannot be rejected. The intra-class correlation is 0.2824, which means that 28.24% of the remaining variance can be explained by differences between countries.
The control variables have a similar relationship with the dependent variable as in Model 4. Again, on the country-level, population, land area and GNI per capita have no significant relationship with the dependent variable. The HIV prevalence rate has a significant positive relationship. On the individual level, the effects are the same too.
The sixth model shows the relationship if both Freedom House and the Freedom of Press are included in the same model. Again, both Freedom House and the freedom of the press do not have a significant effect on ‘Know a place to get HIV test’. The effects of the country-level and the individual level variables stay similar to the previous models. The intra-class correlation is 0.2888, which means that 28.88% of the remaining variance can be explained by differences between countries.
Tab le 3. M u lti -l eve l r egr es si on an al ys is on “ K n ow a p lac e to ge t H IV te st ?” M ode l 4 M ode l 5 M ode l 6 b Int erc ept 0. 000 (5 .001) 0. 000 (5 .009) 0. 000 (4 .998) F re edom H ous e (O ne ye ar pri or t o s urve y) 1. 025 (0 .113) 0. 880 (0 .205) F re edom P re ss (O ne ye ar pri or t o s urve y) 1. 006 (0 .012) 1. 016 (0 .021) Count ry -l ev el v ar iabl es P opul at ion (N at ura l L og ) 1. 709 (0 .316) 1. 698 (0 .313) 1. 686 (0 .314) L and A re a (N at ura l L og ) 0. 738 (0 .242) 0. 746 (0 .236) 0. 759 (0 .243) G N I pe r c api ta , P P P (N at ura l L og ) 1. 122 (0 .320) 1. 136 (0 .317) 1. 151 (0 .316) H IV pre va le nc e (%) 1. 120** (0 .033) 1. 119** (0 .033) 1. 117** (0 .033) Indi vi dual -l ev el v ar iabl es A ge 1. 020*** (0 .002) 1. 020*** (0 .002) 1. 020*** (0 .002) E duc at ion i n ye ars 1. 164*** (0 .009) 1. 164*** (0 .009) 1. 164*** (0 .009) Re li gi on (Re f. N ot re li gi ous ) 1. 433*** (0 .096) 1. 433*** (0 .096) 1. 433*** (0 .096)
G ende r (Re f. M al e) 1. 185 (0 .088) 1. 185 (0 .088) 1. 185 (0 .088) M ari ta l S ta tus (Re f. N ot m arri ed) 1. 880*** (0 .079) 1. 880*** (0 .079) 1. 880*** (0 .079) L it era cy (Re f. Ca n’t re ad) 1. 067 1. 067 1. 067 (0 .041) (0 .041) (0 .041) P la ce of re si de nc e (Re f. U rba n) 0. 584*** (0 .056) 0. 584*** (0 .056) 0. 584*** (0 .056) N um be r of s exua l pa rt ne rs (e xc l. s pous e) 1. 348*** (0 .067) 1. 348*** (0 .067) 1. 348*** (0 .067) K now le dge of H IV tra ns m is si on 4. 329*** (0 .109) 4. 329*** (0 .109) 4. 329*** (0 .109) Re si dua l V ari anc e 3. 29 0 a 3. 29 0 3. 29 0 Int erc ept V ari anc e 1. 303 1. 295 1. 336 ICC (%) 28 .36% 28 .24% 28 .88% No te : M ul ti-Le ve l Logi st ic R egr es si on odds r at io coe ffi ci ent s w ith st andar d er ror s be tw ee n br ac ke ts . D ep en de nt v ar ia ble is ‘ D o y ou know a pl ace for a H IV tes t?” w ith the re fe re nc e ca te go ry : N o. *** p < 0 .001, ** p < 0 .01, * p < 0 .05 a " # (Sni $ jde rs & B os ke r, 1999 ) b Th e mo de l i s te st ed o n mu lti co lli ne ar ity , S ee a pp en di x 4
5. Conclusion
During the same period, (re-)democratization and the HIV/AIDS epidemic were on the rise in Sub-Saharan Africa. Both phenomena have strongly affected the region but the relationship between them has faced little research. In the light of the democratic dividend theory, this paper contributed to addressing this gap with a multi-level cross-national research focused on democracy and press freedom and their effects on citizens’ knowledge of HIV/AIDS and the provision of HIV-testing facilities.
The cross-national analyses showed that there is no significant relationship between democracy and HIV knowledge. Moreover, the relationship between freedom of the press and knowledge of HIV also turned out not to be significant. The relationship between democracy and the provision of HIV-testing facilities as well as between press freedom and testing facilities were similarly not significant.
These results lead to an important question: does democracy bring any merits regarding HIV/AIDS? According to this analysis, this is not the case. Both the respondent’s knowledge of HIV/AIDS and the provision of HIV-testing facilities are not affected by democracy. The level of democracy therefore does not appear to influence HIV/AIDS policies. In this case, one could question the public healthcare policies in democratic Sub-Saharan African countries and, in line with the democratic dividend theory, the legitimacy of the regimes as well.
A possible explanation for the non-result lies with the electorate. Indeed, do they actually demand more HIV/AIDS policies? To be able to reject the legitimacy of the democratic regimes in Sub-Saharan Africa, this should be examined. Taboos and stigmas regarding the illness still play a part in the continent (Hess & Mckinney, 2007; Mbonu, Van Den Borne, & De Vries, 2009). According to previous research, the ideas of AIDS being associated with certain ethnicities, religious believes, immoral sexual behavior and being an illness fabricated by the West are still present in the contemporary Sub-Saharan African countries (Hess & Mckinney, 2007; Mbonu et al., 2009). These stigmas lead to discrimination of those carrying HIV and denial of the illness (Mbonu et al., 2009). Therefore, the electorate could be less demanding for HIV/AIDS policies, even when living in a full democracy. More research on public healthcare is necessary to be able to reject the legitimacy of the democratic regimes in Sub-Saharan Africa.
Besides the non-results of democracy, the freedom of the press also lacks significant
relationships. As mentioned before, having a democracy does not equate to having freedom of the press. However, similar to democracy, freedom of the press does not affect the provision of HIV- testing facilities or the respondent’s knowledge of HIV/AIDS. Apparently, freedom of the press does not influence government policies in such way that it will lead to more HIV-testing services or the provision of more HIV/AIDS-related information.
However, this analysis is only focused on a small part of the African healthcare system, it is remarkable that democracies do not bring any improvement regarding HIV/AIDS policies. According to the existing theories, democracies should bring an increase in living conditions (Cheeseman, 2015). Providing HIV-related information and HIV-testing facilities is an important part of fighting the epidemic (World Health Organization, 2019). The improvement of living conditions is apparently not the case concerning HIV/AIDS policies. Further research should reveal what exactly is the cause of this lack of effect. Moreover, research should be conducted to ascertain whether the lack of a relationship only occurs in Sub-Saharan Africa, or whether this is a more global phenomenon.
Bibliography
Adsera, A., Boix, C., & Payne, M. (2003). Are You Being Served? Political Accountability and Quality of Government. Journal of Law, Economics, and Organization, 19(2), 445-490. https://doi.org/10.1093/jleo/ewg017
Ake, C. (1993). The Unique Case of African Democracy. International Affairs (Royal
Institute of International Affairs), 69(2), 239-244. https://doi.org/10.2307/2621592
Avert. (2018, October 4). How HIV infects the body and the lifecycle of HIV. Retrieved May 7, 2019, from https://www.avert.org/about-hiv-aids/how-infects-body
Avert. (2019-a, March 21). Symptoms and stages of HIV infection. Retrieved May 7, 2019, from https://www.avert.org/about-hiv-aids/symptoms-stages
Avert. (2019-b, May 21). Women and girls, HIV and AIDS. Retrieved June 9, 2019, from https://www.avert.org/professionals/hiv-social-issues/key-affected-
populations/women
BBC News. (2006, April 5). SA's Zuma 'showered to avoid HIV'. Retrieved May 13, 2019, from http://news.bbc.co.uk/2/hi/africa/4879822.stm
Besley, T., & Kudamatsu, M. (2006). Health and Democracy. American Economic Review,
96(2), 313–318. https://doi.org/10.1257/000282806777212053
Boseley, S. (2008, November 26). Mbeki Aids denial 'caused 300,000 deaths'. Retrieved May 13, 2019, from https://www.theguardian.com/world/2008/nov/26/aids-south-africa
Buvé, A., Bishikwabo-Nsarhaza, K., & Mutangadure, G. (2002). The spread and effect of HIV-1 infection in sub-Saharan Africa. The Lancet, 359(9322), 20HIV-1HIV-1-20HIV-17. https://doi.org/10.1016/S0140-6736(02)08823-2
Bueno de Mesquita, B., A. Smith, R. M. Siverson, and J. D. Morrow. 2003. The Logic of
Bratton, M., Mattes, R., & Gyimah-Boadi, E. (2005). Modeling Attitudes to Reform. Public
Opinion, Democracy and Market Reform in Africa (pp. 271-294). Cambridge:
Cambridge University Press.
Bratton, M., & Van der Walle, N. (1997). Democratic Experiments in Africa: Regime
Transitions in Comparative Perspective. Cambridge: Cambridge University Press.
Cheeseman, N. (2015). Democracy in Africa: successes, failures, and the struggle for political
reform. Cambridge University Press.
Deacon, R. T. (2009). Public good provision under dictatorship and democracy. Public
choice, 139(1-2), 241-262. https://doi.org/ 10.1007/s11127-008-9391-x
DHS. (2008). Demographic and Health Survey Model Woman's Questionnaire (Phase 5). Retrieved from https://www.dhsprogram.com/pubs/pdf/DHSQ5/DHS5-Woman's- QRE-22-Aug-2008.pdf
DHS. (2018a). Demographic and Health Survey Model Woman's Questionnaire (Phase 7). Retrieved from https://dhsprogram.com/pubs/pdf/DHSQ7/DHS7-Womans-QRE-EN- 17Dec2018-DHSQ7.pdf
DHS. (2018b). Guide to DHS statistics (DHS-7). Retrieved from
https://dhsprogram.com/pubs/pdf/DHSG1/Guide_to_DHS_Statistics_DHS-7.pdf
DHS. (n.d.-a). The DHS Program - DHS Methodology. Retrieved on April 20, 2019, from https://www.dhsprogram.com/What-We-Do/Survey-Types/DHS-Methodology.cfm
DHS. (n.d.-b). The DHS Program - DHS Questionnaires. Retrieved on April 20, 2019, from https://www.dhsprogram.com/What-We-Do/Survey-Types/DHS-Questionnaires.cfm
D'Souza, F. (1994). Democracy as a Cure for Famine. Journal of peace research, 31(4), 369-373. https://doi.org/10.1177%2F0022343394031004001
Duval Smith, A. (2011, September 21). The President who claims he can cure Aids on
Mondays. Retrieved May 13, 2019, from
https://www.independent.co.uk/news/world/africa/the-president-who-claims-he-can-cure-aids-on-mondays-434797.html
Franco, Á., Álvarez-Dardet, C., & Ruiz, M. T. (2004). Effect of democracy on health:
ecological study. BMJ, 329(7480), 1421–1423.
https://doi.org/10.1136/bmj.329.7480.1421
Freedom House. (2017a, May 15). Freedom of the Press 2017 Methodology. Retrieved April 30, 2019, from https://freedomhouse.org/report/freedom-press-2017-methodology
Freedom House. (2017b, November 9). Freedom of the Press [Dataset]. Retrieved on April 20, 2019, from https://freedomhouse.org/sites/default/files/FOTP1980-
FOTP2017_Public-Data.xlsx
Freedom House. (2018, May 8). Country and Territory Ratings and Statuses, 1973-2018 [Dataset]. Retrieved on April 20, 2019, from
https://freedomhouse.org/sites/default/files/Country%20and%20Territory%20Ratings %20and%20Statuses%20FIW1973-2018.xlsx
Freedom House. (2019, January 31). Methodology: Freedom in the World 2018. Retrieved May 2, 2019, from https://freedomhouse.org/report/methodology-freedom-world-2018
Gwaambuka, T. (2016, October 1). African Presidents and Politicians and Their Crazy Ideas
About AIDS. Retrieved May 13, 2019, from
https://www.africanexponent.com/post/7548-african-politicians-and-their-crazy-ideas-about-aids
Harsch, E. (n.d.). AIDS deaths are declining, reports UN | Africa Renewal. Retrieved May 13, 2019, from https://www.un.org/africarenewal/magazine/january-2008/aids-deaths-are-declining-reports-un
Hess, R., & McKinney, D. (2007). Fatalism and HIV/AIDS Beliefs in Rural Mali, West Africa. Journal of Nursing Scholarship, 39(2), 113–118.
https://doi.org/10.1111/j.1547-5069.2007.00155.x
Hiv.gov. (2018, September 25). How Is HIV Transmitted? Retrieved May 7, 2019, from https://www.hiv.gov/hiv-basics/overview/about-hiv-and-aids/how-is-hiv-transmitted
Human Rights Watch. (2014, May 14). Uganda: Deeply Flawed HIV Bill Approved. Retrieved May 13, 2019, from https://www.hrw.org/news/2014/05/13/uganda-deeply-flawed-hiv-bill-approved
International Partnership for Microbidices. (n.d.-a). How HIV Infects a Cell. Retrieved May 7, 2019, from https://www.ipmglobal.org/how-hiv-infects-cell
International Partnership for Microbicides. (n.d.-b). Women & HIV/AIDS. Retrieved June 9, 2019, from https://www.ipmglobal.org/women-hiv-aids
Justesen, M. K. (2012). Democracy, dictatorship, and disease: Political regimes and HIV/AIDS.
European Journal of Political Economy, 28(3), 373-389.
https://doi.org/10.1016/j.ejpoleco.2012.02.001
Lake, D. A., & Baum, M. A. (2001). The invisible hand of democracy: political control and the provision of public services. Comparative political studies, 34(6), 587-621.
https://doi.org/10.1177%2F0010414001034006001
Kagaayi, J., & Serwadda, D. (2016). The History of the HIV/AIDS Epidemic in
Africa. Current HIV/AIDS Reports, 13(4), 187–193. https://doi.org/10.1007/s11904- 016-0318-8
Mail & Guardian. (2011, November 21). Africa sees dramatic improvement in HIV treatment. Retrieved May 13, 2019, from https://mg.co.za/article/2011-11-21-africa-sees-dramatic-improvement-in-hiv-treatment
Marsaudon, A., & Thuilliez, J. (in press). Does democracy reduce the HIV epidemic? Evidence from Kenya. HAL, 1–16. Retrieved from https://halshs.archives-ouvertes.fr/halshs-01400833/document
Masaki, T., & Van de Walle, N. (2014). The impact of democracy on economic growth in sub-Saharan Africa, 1982 - 2012, WIDER Working Paper, No. 2014/057, The United Nations University World Institute for Development Economics Research (UNU- WIDER), Helsinki. https://www.wider.unu.edu/publication/impact-democracy- economic-growth-sub-saharan-africa-1982-2012
Mbonu, N., van den Borne, B., & De Vries, N. (2009). Stigma of People with HIV/AIDS in Sub-Saharan Africa: A Literature Review. Journal of Tropical
Medicine, 2009(145891), 3–16. https://doi.org/10.1155/2009/145891
Mutume, G. (n.d.). African women battle for equality. Retrieved April 21, 2019, from https://www.un.org/africarenewal/magazine/july-2005/african-women-battle-equality
New Vision. (2004, May 19). Museveni down plays condoms. Retrieved May 13, 2019, from https://www.newvision.co.ug/new_vision/news/1096671/museveni-plays-condoms
Przeworski, A., & Cheibub, J. (1999). Democracy, Elections, and Accountability for Economic Outcomes. In A. Przeworski, S. Stokes, & B. Manin (Reds.), Democracy,
Accountability, and Representation (pp. 222-249). Cambridge, Great Brittain:
Cambridge University Press.
Quinn, T., Mann, J., Curran, J., & Piot, P. (1986). AIDS in Africa: an epidemiologic paradigm. Science, 234(4779), 955–963. https://doi.org/10.1126/science.3022379
Rosenzweig, S. C. (2015). Does electoral competition affect public goods provision in dominant-party regimes? Evidence from Tanzania. Electoral Studies, 39, 72–84. https://doi.org/10.1016/j.electstud.2015.04.004
Ross, M. (2006). Is democracy good for the poor?. American Journal of Political
Science, 50(4), 860-874. https://doi.org/10.1111/j.1540-5907.2006.00220.x
Ruger, J. (2005). Democracy and health. QJM, 98(4), 299–304.
https://doi.org/10.1093/qjmed/hci042
Safaei, J. (2006). Is democracy good for health?. International Journal of Health Services, 36(4), 767-786.
Schoepf, B. G. (2004). Museveni's Other War: Condoms in Uganda. Review of African
Political Economy, 31(100), 372–376. https://www-jstor-
org.ezproxy.leidenuniv.nl:2443/stable/4006901
Sen, A. (1982). Poverty and famines: an essay on entitlement and deprivation. Oxford university press.
Sen, A. (1994). Freedom and Needs. New Republic. Vol 210 (2/3), pp. 31-38.
http://web.b.ebscohost.com.ezproxy.leidenuniv.nl:2048/ehost/detail/detail?vid=0&sid =abb547fe-a273-4f9a-a456-
aba79560f98d%40sessionmgr120&bdata=JnNpdGU9ZWhvc3QtbGl2ZQ%3d%3d#db =aph&AN=9312307600
Sen, A. (1999). Development as freedom. New York: Alfred A. Knopf.
Sharp, P. M., Bailes, E., Chaudhuri, R. R., Rodenburg, C. M., Santiago, M. O., & Hahn, B. H. (2001). The origins of acquired immune deficiency syndrome viruses: where and when? Philosophical Transactions of the Royal Society of London. Series B: Biological
Sciences, 356(1410), 867–876. https://doi.org/10.1098/rstb.2001.0863
Sharp, P. M., & Hahn, B. H. (2011). Origins of HIV and the AIDS Pandemic. Cold Spring
Harbor Perspectives in Medicine, 1(1), 1–22.
Snijders, T. A., & Bosker, R. J. (1999). An introduction to basic and advanced multilevel modeling. Sage, London. WONG, GY, y MASON, WM (1985): The Hierarchical
Logistic Regression. Model for Multilevel Analysis, Journal of the American Statistical Association, 80(5), 13-524.
Söderlund, N., Lavis, J., Broomberg, J., & Mills, A. (1993). The costs of HIV prevention strategies in developing countries. Bulletin of the World Health Organization, 71(5),
595–604. Retrieved from
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2393468/pdf/bullwho00038-0111.pdf
Treatment Action Campaign. (n.d.). About TAC. Geraadpleegd op 29 april 2019, van https://tac.org.za/category/about/
UN AIDS. (2006). Keeping the Promise: An Agenda for Action on Women and AIDS. Retrieved from
http://data.unaids.org/pub/booklet/2006/20060530_fs_keeping_promise_en.pdf
UN AIDS. (2010). UNAIDS report on the global AIDS epidemic . Retrieved from http://files.unaids.org/en/media/unaids/contentassets/documents/unaidspublicatio n/2010/20101123_globalreport_en%5B1%5D.pdf
UN AIDS. (2018). FACT SHEET – WORLD AIDS DAY 2018. Retrieved from http://www.unaids.org/sites/default/files/media_asset/UNAIDS_FactSheet_en.pdf
Van der Windt, P., & Vandoros, S. (2017). Democracy and health: Evidence from within- country heterogeneity in the Congo. Social Science & Medicine, 194, 10–16. https://doi.org/10.1016/j.socscimed.2017.10.008
Volkskrant. (2006, April 6). Zuma nam douche om aids te voorkomen. Retrieved May 13, 2019, from https://www.volkskrant.nl/wetenschap/zuma-nam-douche-om-aids-te-voorkomen~b720027b/
Wigley, S., & Akkoyunlu-Wigley, A. (2011). The Impact of Regime Type on Health: Does Redistribution Explain Everything? World Politics, 63(04), 647–677. https://doi.org/10.1017/s0043887111000177
World Bank Indicators. (n.d.). World Development Indicators | DataBank [Dataset]. Retrieved on April 20, 2019, from https://databank.worldbank.org/data/reports.aspx?source=2
World Health Organization. (2018). HIV/AIDS surveillance in Europe 2018. Retrieved from https://ecdc.europa.eu/sites/portal/files/documents/hiv-aids-surveillance-europe- 2018.pdf
World Health Organization. (2019). Why the HIV epidemic is not over. Retrieved May 7, 2019, from https://www.who.int/hiv-aids/latest-news-and-events/why-the-hiv-epidemic-is-not-over
Zou, J., Yamanaka, Y., John, M., Watt, M., Ostermann, J., & Thielman, N. (2009). Religion and HIV in Tanzania: influence of religious beliefs on HIV stigma, disclosure, and treatment attitudes. BMC Public Health, 9(1), 1-12. https://doi.org/10.1186/1471-2458- 9-75
Appendix 1
A. Variable Description
Dependent Variables
1. “Ever heard of AIDS?”
This variable is binominal and derived from the DHS survey question 1001: “Now I would like
to talk about something else. Have you ever heard of HIV or AIDS?” (DHS, 2018a, p.66) The
two answer options are: ‘yes’ and ‘no’. All other answers are coded as missing. If the respondent answers this question with ‘yes’, more HIV/AIDS related questions are being asked. If the respondent answers with ‘no’, no further HIV-related questions are being asked.
2. “Know a place to get HIV test?”
This variable is again derived from the DHS survey: “Do you know of a place where people
can go to get an HIV test?” (DHS, 2018a, p.68). Again, this variable is binominal with the only
answer options being ‘yes’ and ‘no’. This question can only be asked if the respondent answered ‘yes’ on the question if they ever heard of HIV or AIDS. Therefore, the fourth, fifth and sixth model have less respondents than the first, second and third.
Independent variables
1. Freedom House Index (FH_0 and FH_1)
Freedom House Index is derived from the Freedom House (2018). Freedom House rates every country on a 1 (most) to 7 (least) scale of democracy. This rating is made up by two different ratings: Civil Liberties (1 to 7) and Political Rights (1 to 7). The mean of these two ratings make up the Freedom House rating. Every year, Freedom House publishes a new rating for the year before. The rating is a result of 25 methodological questions which make up the final score. The analysis is done by analysts and experts of different institutions (2019).
FH_0 is the freedom house rating on a scale of 1 to 7 in the year of the survey. FH_1 is the freedom house rating on a scale of 1 to 7 in the year prior to the survey
2. Freedom of the Press Index (Press_0 and Press_1)
Freedom of the Press index is derived from the Freedom House (2017b) too. Freedom of the press has a scale of 0 (most) to 100 (least) free, which is prepared by means of 23 methodological questions. These methodological questions are divided in three categories:
