THE KINGDOM OF SWAZILAND NATIONAL EMERGENCY RESPONSE COUNCIL ON HIV/AIDS Swaziland Hearsay Ethnography Study Final report November 2011 THE WORLD BANK © International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Internet: www.worldbank.org; Telephone: 202 473 1000 This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or other partner institutions or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 Unported licence (CC BY 3.0 IGO) https://creativecommons.org/licenses/by/3.0/igo/. Under the Creative Commons Attribution license, you are free to copy, distribute and adapt this work, including for commercial purposes, under the following conditions: Attribution – Please cite the work as follows: The World Bank. 2011. Swaziland Hearsay Ethnography Study. Washington DC: World Bank. License: Creative Commons Attribution CC BY 3.0 IGO. Translations – If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in its translation. All queries on rights and licenses should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington DC, 20433, USA; fax: 202-522-2625; email: pubrights@worldbank.org. TABLE OF CONTENTS ACKNOWLEDGEMENTS ...........................................................................................................................i ACRONYMS............................................................................................................................................ ii EXECUTIVE SUMMARY ...................................................................................................................... iii SECTION 1. BACKGROUND AND RATIONALE FOR THE STUDY ................................................................1 SECTION 2. DATA AND METHODS .......................................................................................................... 2 2.1 Selection of sites ..........................................................................................................................2 2.2 Selection and characteristics of the ethnographers ..................................................................... 2 2.3 Ethical approval of the study ........................................................................................................ 3 2.4 Training........................................................................................................................................ 3 2.5 Data collection, coding and analysis .............................................................................................3 2.6 Quality assurance of the data ....................................................................................................... 4 2.7 Strengths and limitations of the data ...........................................................................................4 SECTION 3. RESULTS ..............................................................................................................................6 3.1 Typology of sexual partnerships .................................................................................................. 6 3.2 Talk about casual partnerships ..................................................................................................... 7 3.3 Motivations for partnerships ...................................................................................................... 11 3.4 Dynamics of sexual behaviour .................................................................................................... 15 3.5 Norms about relationships ......................................................................................................... 16 3.6 Talk of marriage ........................................................................................................................ 19 3.7 The 4 C’s: Circumcision, Concurrency, Condoms and Clinics ....................................................... 21 SECTION 4. CONCLUSIONS .................................................................................................................. 31 4.1 The value of talk......................................................................................................................... 31 4.2 Types of partnerships ................................................................................................................. 31 4.3 Motivations for partnerships ...................................................................................................... 32 4.4 Cultural scripts and norms.......................................................................................................... 32 4.5 Male circumcision ...................................................................................................................... 33 4.6 Sexual concurrency .................................................................................................................... 33 4.7 Misconceptions relating to HIV/AIDS.......................................................................................... 33 4.8 In conclusion .............................................................................................................................. 34 BIBLIOGRAPHY..................................................................................................................................... 35 Annexes Annex 1: Data and Methods .............................................................................................................. 38 Annex 2: Hearsay Ethnography Training Notes for Ethnography Recruits & Suggestions for Trainers 44 Annex 3: Coding protocols (Heinrich Crouse) .................................................................................... 46 List of figures Figure 1: Attitudes towards intergenerational sex ................................................................................. 10 Figure 2: Attitudes towards intergenerational sex by gender and age ................................................... 10 Figure 3: Attitudes against intergenerational sex by types of partners................................................... 11 Figure 4: Frequencies of positive and negative comments about marriage ........................................... 20 Figure 5: Attitudes towards marriage by gender and age ...................................................................... 21 Figure 6: Frequency of comments for and against male circumcision ................................................... 24 Figure 7: Arguments against male circumcision.................................................................................... 25 Figure 8: Arguments for male circumcision........................................................................................... 26 Figure 9: Negative comments about concurrency ................................................................................. 27 Figure 10: Estimated duration of concurrent partnerships ..................................................................... 28 Figure 11: Frequency of comments for and against condom use .......................................................... 29 Figure 12: Arguments for condom use ................................................................................................. 29 Figure 13: Reasons for not using condoms .......................................................................................... 30 ACKNOWLEDGEMENTS This report presents the results of qualitative research commissioned by the World Bank in collaboration with the National Emergency Response Council on HIV and AIDS, Swaziland (NERCHA). NERCHA would like to thank the following stakeholders for their inputs and invaluable contributions: • NERCHA senior management for their support of this groundbreaking type of research work. • The research team, led by Susan Cotts Watkins (World Bank consultant), with Swaziland-based hearsay ethnography field workers collecting data over a six month period. • The assistant researchers, Lucas Jele and Xoli Sereo from NERCHA. • Heinrich Crouse (World Bank consultant), responsible for data coding and analysis, and supporting report writing. • The local coders, Sifiso Sithole, Ruth Mkhonta, Siphiwe Thwala, Hlobsile Nkambule-Bhembe all from the University of Swaziland and Thabile Maripe an independent consultant • The University of Swaziland for allowing their staff to participate in the ground breaking research Also, special thanks go to the HIV Prevention Team at NERCHA for their leadership in relation to this study, the monitoring and evaluation unit at NERCHA, members of the M&E technical working group (TWG) and to all the stakeholders in Swaziland for their feedback on the findings and data interpretations. Following presentations of an earlier version of this report to NERCHA and other stakeholders, and based on comments made at these presentations, the draft report was revised, using additional data analysis on an expanded range of topics and increased detail of the coding. Thanks also go to Stephane Helleringer (Columbia University, New York) for his comments on the draft report, and to Lusweti Institute for Health Development Communication, for support with the hearsay ethnography data collection and the financial administration of the research. Members of the World Bank HIV prevention team -- Marelize Gorgens, Nicole Fraser-Hurt and Susanne Stromdahl – provided comments and supported the finalisation of the report. NERCHA wants to thank the World Bank for availing the resources for the research study and would also like to thank PEPFAR and PSI who contributed additional resources to support the study with regards to the engagement of the local coders. i ACRONYMS AIDS Acquired Immune Deficiency Syndrome CSW Commercial Sex Worker HIV Human Immunodeficiency Virus MC Male Circumcision MCP Multiple and/or Concurrent Partners MOT Modes of Transmission NERCHA National Emergency Response Council on HIV and AIDS STI Sexually Transmitted Infection ii EXECUTIVE SUMMARY This study used an innovative methodology of qualitative data collection - hearsay ethnography. The methodology gives insights into sexual and societal norms and cultural scripts held by community members. Trained ethnographers listened to public conversations about HIV-related topics which occurred in public and could be noted down in journals. The ethnographers were not to ask questions or direct the conversation in any way. The research was carried out in several places in Swaziland in order to capture some of the diversity in the country: Ludzeludze, Bhuleni, Siphofaneni and Mathandele. Two of the sites, Buhleni and Siphofaneni, are “rural growth points”, i.e. rapidly growing areas. Ludzeludze, in the vicinity of Manzini, has considerable urban influence, and Mathandele is under the jurisdiction of the Nhlangano Town Council and considered urban. The research found that there is a great deal of talk about all types of sexual partnerships, from marriage to multiple sexual partners to commercial sex work. The most common topic is non-marital partnerships of all types. The conversations are critical of non-marital partnerships, and are often studded with insults about those who violate community norms about proper sexual behaviour. The main topics covered in these community conversations have been coded and are described below: People’s public views about casual sex. Of the total mentions of casual sex in the overheard and documented conversations, about one-quarter were positive: almost all of these were made by young men. No older women or older men claimed that casual sex was a “good behaviour”. The most frequent criticisms are that casual sex is improper, even immoral, behaviour and that it leads to AIDS. However, it has to be remembered that this methodology documents public discussions and that people – despite showing disapproval about a behaviour in public – may still secretly adopt the behaviour for some perceived benefits. The Swaziland HIV Prevention Response and Modes of Transmission Analysis (2009) describes the practice of multiple, long-term concurrent partnerships as “perceived to be unacceptable, yet tacitly acknowledged and tolerated in Swazi society” (vi). Although it is difficult to isolate talk about long-term multiple and/or concurrent partners (MCP) in the ethnographic data, all types of non-marital sexual partnerships are publicly criticized. People’s public views about marriage. One of the central themes in the data analysis was the norm and cultural scripts around marriage. Is marriage desirable? Is it something that people want, or is it something that is removed and distant? Numerically the reality in Swaziland, as well as in some neighbouring countries, is that marriage rates are exceptionally low. One of the surprising findings of this study is the continuing publicly-expressed, positive norms about marriage. Of all the excerpts from the journals on marriage, twice as many were positive about marriage as were negative. Some of the reasons for marrying suggest that it is aspirational, e.g. the speaker says that he or she wants the love and trust of marriage, to make her parents happy, to have a white wedding, “I want to choose my own wife”. Other comments suggest more urgency, e.g. “I have children already and I want to marry the father of the child”, “the time for her getting married is getting closer”; that the person does not approve of sex outside of marriage and is thus waiting for, and presumably expecting, marriage. Overall, it can be concluded that although marriage rates are very low, there is a notable desire for marriage and the benefits marriage may confer. Publicly-expressed motivations for casual partnerships. Those who tell stories about people who fail to fulfill the norms primarily locate the blame on four motivations for improper behaviour: Love, money, revenge and sexual desire. The refined coding of the journal data permitted further analyses which show that the relative frequency of these motivations varies by the gender of the speaker as well as the gender of the person whose behaviour is the topic of conversation. This suggests that all data on sexual behaviour should be seen through a gender lens: conclusions based only on data from one gender would not accurately describe the perceptions of both genders. Detailed coding to capture potential variations by age, education and employment status did not reveal any patterns, but there was iii not sufficient data in the journals to support a definitive analysis of these characteristics. When women are telling stories about other women, sex—either for enjoyment or as a physical need-- is the most frequently-attributed motivation, followed by the desire, or need, for money, with a search for love or a desire for revenge being markedly less frequent. In contrast, when men are speaking about women, by far the most frequent motivation for casual partnerships is perceived to be money: women are primarily seen as mercenary rather than vulnerable. Women’s sexual desire is the second most common motivation attributed to women by men. In the conversations, women’s unemployment was very rarely mentioned. However, the dominant characterizations of women in the journals, by both male and female participants, was that women actively seek men with money, not that they are driven to sex for survival (it was probably taken for granted by participants in the conversations that employment would improve women’s economic status, and benefit especially the very poor). This characterization, along with the strength of the other motivations for sexual partnerships attributed to women, suggests that even if opportunities for female labour force participation were more abundant and lucrative, this would be unlikely to have a major impact on women’s sexual behaviour. In respect to men, it is widely believed that the ability of a man to have partners depends on his income. Unemployment would therefore decrease a man’s success in obtaining sexual partners. The typology of sexual partnerships identified a total of 12 types of partnerships – the married couple (without other partners) and 11 partnership types with casual and non-marital partners involved. The partnerships were categorized by marital status (married or single), by whether the partner who was mentioned was a steady or a casual partner, and by whether one partner had multiple casual partners. The use of ethnographic data on informal conversations does not permit assessing the relative frequency of the various types of partnerships, but rather the frequency distribution of comments about partnerships. The type of casual partnership most likely to elicit criticism are partnerships where at least one of the partner has multiple casual partners: “She is food for the nation”, “Men are the ones spreading the disease by failing to control sexual urges when they are drunk and not using condoms”, “He has first borns all over”. Little was recorded in the journals about the venues for sexual activity, or even where the proposals that initiated it were made. Of the comments that specified a venue, almost all were in the journals of three male ethnographers in Buhleni. By their account, the two most common venues were drinking places and at ceremonies such as weddings and the Marula festival. For the purposes of developing the typology of sexual partnerships, it was necessary to consider each type of partnership as distinct from every other type. This gives the impression of rigidity and stability. However, it appears that participants in conversations about sexual partnerships see the boundaries between partnership types as porous: a casual relationship can become a live-in relationship, which in turn has the potential to become a marriage. Correspondingly, a live-in relationship can devolve into a sporadic casual relationship, or end entirely. Although it was hoped for, the study did not provide insight into the temporal dynamics of sexual behaviour in Swazi communities i.e. any changes in sexual behaviour over time. There was very little comment on this aspect in the ethnographies, since the conversations were essentially about what is happening now to people in their community (nor did the literature review on sexual behaviour in Swaziland in the past provide much insight into temporal dynamics of sexual behaviours in Swaziland). It does, however, suggest that at least as far back as the 1930s, and perhaps into the pre-colonial period, non-marital partnerships were common, and that declines in male unemployment lead to a concentration of marriage in the middle class. However, should a repeat of this study be carried out, temporal dynamics may be visible. Sexual concurrency. In conversations about sexual partnerships, it was rarely clear whether it was an overlapping i.e. concurrent partnership or a partnership with multiple sequential partners. People did not seem to make an obvious distinction, and when talk was of multiple partnership (which may have been concurrent or serial), more than half of the respective comments did not mention HIV or AIDS at all, but iv rather were about other aspects of partnerships such as children, rights of men to have multiple partners, pride in multiple partners, or complaints about a cheating partner. Even when there was talk of “sharing” partners, it was often not clear whether the sharing was concurrent or sequential. Awareness of male circumcision. The study showed that the efforts to promote male circumcision have generated vigorous debate, particularly among the young men who are the target of the campaign. In the documented conversations, there are more comments disapproving of male circumcision than approving of it. Two of the three major arguments against circumcision are concerns about its impact on sexual prowess: injury, which is related to the fear that performance will be hindered, and direct comments expressing concern about the effect of circumcision on prowess. HIV-related misconceptions. While available survey data identified several misconceptions in respect to HIV and AIDS, the conversations as documented in the journals provide no evidence that these matter for sexual behaviour. In talk about how to avoid infection, avoiding mosquitoes is never mentioned, and sharing plates is rarely mentioned. It thus appears that reports based on survey data that feature these misconceptions are inconsistent with local perceptions about the location of the risk of AIDS: in the journals, it is from non-marital partners primarily, and secondarily from transmission within marriage. The journals are valuable in uncovering pervasive misconceptions that have not been addressed in surveys, which points to the advantages of listening to people talking to each other in their own communities. It is clear from the conversations that members of local communities do not appreciate the possibility that a married couple might be HIV sero-discordant, which has implications for HIV testing; they believe that the duration from infection to symptoms is shorter than it is in reality and this affects risk perception and partner selection ; and they have only a vague understanding about the correspondence of HIV prevalence and age, with particular implications for the risk young women incur in relationships with much older men. Lack of accurate information on the epidemiology of HIV may be a significant barrier to prevention: thus, providing such information has the potential to be effective in reducing HIV transmission and support other messaging relating directly to behaviour change such as partner reduction messages. In summary, the following key messages can be retained from this study: • Hearsay ethnography provides an opportunity to learn what ordinary people living in the communities in Swaziland say to each other about relationships between the sexes and norms attached to different sexual behaviours. It is therefore an additional data source which can complement data from interviews of informants and from focus group discussions, in order to inform communication aspects of HIV prevention programmes. • Conversations about sexual behaviour and partnerships among Swazi men and women are frequent. People perceive AIDS still as dangerous. Fatalistic attitudes seem to be infrequent. • Twelve distinct relationship types could be defined, ranging from marriage without the presence of other partners, to relationships with commercial sex workers. Two types of behaviour are subject to particularly severe criticism: men and women who have multiple partners, and intergenerational relationships. The former are seen as particularly risky and the latter violate local norms. Overall, risky sexual behaviour is perceived as “bad” on a number of dimensions. • Swazi people talk critically about extra-marital relationships. According to the comments recorded in the study, extramarital sex is transactional: it involves a transaction between a man with money and a woman who wants money. • One-night stands and buying sex from commercial sex workers seem relatively rare based on public conversations (or these are not readily admitted to in public conversations). v • In the documented conversations, no distinction is apparent when people talk about serial or concurrent partners in terms of risks entailed or avoidance of such risks. • Young women don’t appear to recognize how risky relationships with older or with wealthy men are, and lack a risk perception informed by epidemiological evidence of increasing HIV prevalence by men’s age and wealth. • Marriage is seen as especially desirable for women, but men are also presented as wanting to be married eventually. Marriage is associated with an expectation of higher standards of acceptable sexual behaviour. • Male circumcision is much talked about, but there is a lack of understanding as to why male circumcision is effective. The issue of sexual prowess dominates men’s conversations, both as a reason for circumcision and as a reason for objecting to it. • The study identified misconceptions held by the population about sero-discordance, the duration from infection to AIDS symptoms, and the correspondence of HIV prevalence and age. A key aspect is for men to understand sero-discordance in couples, since they wrongly believe that the sero-status of their female partner – often assessed within antenatal care - predicts their own sero-status. vi SECTION 1. BACKGROUND AND RATIONALE FOR THE STUDY In 2009, the Kingdom of Swaziland and NERCHA announced the National Strategic Framework for HIV and AIDS 2009-2014. This reported that annual HIV incidence peaked around 1999 at almost 6%. Subsequent estimates that by 2006 incidence was 2.9%, which is expected to decline to 2.18% by 2015 (Kingdom of Swaziland HIV Estimates and Projections Report, 2010, 6). Although it is difficult to isolate the specific causes of the change, it is likely to have been some combination of prevention programmes, spontaneous responses in the communities as people responded to the deaths of their relatives, friends and neighbors (Watkins 2004), and the natural course of an epidemic, i.e. a reduced number of susceptibles who may overall have less risky sexual behaviours than those already infected (Watkins 2004; Garnett, Gregson and Stanecki 2006). To reduce HIV incidence further, continued prevention efforts were deemed to be critical for the national response to HIV. The Swaziland HIV Prevention Response and Modes of Transmission (MOT) study, conducted in 2009, called for a re-orientation of prevention programmes. The report on the MOT study noted that most prevention programmes have focused on youth or the general population, but that people in stable relationships appeared to be a distinct group at risk with a large number of new HIV infections estimated to occur among them. Targeting both married and unmarried individuals in stable relationships therefore seemed a priority. The MOT report noted that neither MCPs and sexual networks nor intergenerational relationships were sufficiently and explicitly addressed in policies and programmes (Government of Swaziland 2009). The National Strategic Framework set as priorities the following: Social and behaviour change communication programmes; Reduction of MCPs among the sexually active population; Increased and comprehensive knowledge of HIV and AIDS; Scaling up of prevention of mother to child transmission of HIV (PMTCT); and Circumcision of HIV negative men. The aim of this research was to collect and analyze an unusual source of data that could assist NERCHA in implementing the framework: People’s public conversations, overheard and recorded by trained ethnographers. The methodology is expected to give some insights into sexual and societal norms and cultural scripts held by community members in Swaziland. 1 SECTION 2. DATA AND METHODS Ethnography is a staple of anthropology. The advantage of ethnography over other forms of qualitative data collection is that it provides information on a community, rather than on individuals, and thus provides information on cultural scripts and norms that are external to individuals and expressed through informal and social interactions (Watkins and Swidler 2009) Hearsay Ethnography uses members of local communities to provide information on public conversations about AIDS. These conversations are focused on gossip, stories about an individual. A narrator tells the story, the participants comment. The underlying assumption of this method is that in conversations, people learn about the views of others in their community. These views then become elements of an individual’s “tool kit”, to be drawn on when making a decision (Swidler 1984). Since the stories are about people much like the participants, they are likely to be particularly influential. Hearsay Ethnography provides unique access to these conversations. This section gives a brief overview on how data were collected and quality assured. Further detail to establish the credibility of the analysis is available in Annexes 1-3. 2.1 Selection of sites The selection of the research sites was made by NERCHA. These sites —Ludzeludze, Bhuleni, Siphofaneni and Mathandele—capture much of the diversity in the country. The sites required careful selection in order to maximise the representativeness and usefulness of the study. The systematic selection process used the following criteria: 1. The site must have a high population density, with the majority of the people in the site resident and not transitory or on short term visits. 2. The site has public spaces where the residents interact with each other, such as shopping centres, sports centres, shebeens or bars, bus stops, churches or markets. 3. Some of the sites must be rural, albeit densely populated, in order to represent the rural Swazi population (according to the population census, about 70% of the population is rural). 4. The four sites must represent the four regions of the country. 2.2 Selection and characteristics of the ethnographers The initial recruitment of potential ethnographers in these sites was done by the clerk of the Kagogo Centre in three sites and by the HIV prevention coordination office of the Nhlangano town council. This was done after provision of the criteria for recruiting the potential ethnographers. There was also a field visit by the research team leader to two of the sites as part of familiarization and assessment of the sites and the potential ethnographers. The clerks were asked to recruit more ethnographers for training than would be needed, since earlier experiences with Hearsay Ethnography in Malawi and Botswana showed that some of those recruited would not be suited to the task and some would drop out. In Swaziland, 33 potential ethnographers were recruited and trained. The ethnographers were a diverse group in terms of their demographic characteristics. This diversity does not mean that the ethnographers were representative of the diversity of their settings or of the general adult population. However, since an individual’s social networks are typically homophilous, it is likely that the diversity of the ethnographers means that their social networks differ. Thus, the conversations they reported in the journals came from many different segments of the population. 2 2.3 Ethical approval of the study The study was conducted following application and approval of the Scientific and Ethics Committee of the Ministry of Health and Social Welfare of Swaziland. 2.4 Training The study team spent three days in each site training the recruits. The training, described in detail in Annex 2, emphasized several key aspects of the methodology: the ethnographer was not to interview, but simply to listen to public conversations about AIDS, defined as conversations that other people could overhear, e.g. on a bus, in a shebeen, at a football game, etc. The ethnographer were instructed not to ask questions or direct the conversation in any way. What this means for the analysis is that the information in the journals is often less complete than one would like. For example, someone might say that “circumcision is bad” or that “marriage is good”, but without providing an explanation. In other data collection methods, an interviewer can ask further questions, or request a clarification, but the ethnographers were told not to do this. The definition of “a conversation about AIDS” was left up to the journalists. Almost invariably, the conversations recorded in the journals had to do with the heterosexual transmission of HIV or deaths from AIDS, although there were also some on antiretrovirals and on orphans. The training emphasized the importance of producing accurate and detailed accounts of the ways people talk to each other in public and in everyday contexts. This would permit NERCHA and others to understand the people share information, the debates they have about how to make sense of the forces acting on their lives, and the stories they tell each other about how life should be lived. An effective way of getting this across was to ask them to think of themselves as video cameras, simply recording in their mind what they heard and saw, and then translating that into a journal account. This made it clear that the research was not about either the ethnographer’s thoughts and behaviour nor the behaviour of those who were speaking in public. Emphasis was also placed on capturing, as much as possible, the exact words of the speakers rather than summarizing, i.e. “He said ‘The more girlfriends the better, that way I can have my choice’” rather than “He talked about having many partners”. During the conversation, no notes were to be taken. Afterwards, the ethnographer could, in a private place, jot down a few notes which would serve as reminders when he/she wrote up the encounter in the evening. Several ethnographers did not do well enough in the training to continue; several did not submit any journals at all or submitted only a few. Two journalists plagiarized and were dismissed; their journals are not used in the analysis. 2.5 Data collection, coding and analysis Some ethnographers produced one journal for each of the six months of the study, some produced only one. The mode was four journals per ethnographer. A total of 106.5 journals were submitted and paid for. Of these, 12 were discarded. This includes all the journals of the two journalists who were dismissed for plagiarism, as well as some journals that the coder and lead consultant thought were not convincing. There were also parts of otherwise-acceptable journals that appeared to have been fabricated: the mode of speech reported in the journal was different from the ethnographer’s other writing, and from the mode of speech in the journals of the other ethnographers. These parts, amounting to approximately three journals, were also ignored in the analysis. There were a total of 992 conversations about AIDS. Some of these were quite long (15 pages), others were brief. The ethnographers had been asked to describe the context in which the conversations occurred, including what the participants were wearing (to provide a sense of their economic status). Often many pages were written on context and clothing; these are not included in the count of the pages 3 of each conversation. The coding scheme was developed by the lead consultant and refined in consultation with Mr Jele and the coding consultant. The coding required intensive reading of the journals. Because of the ambiguities of ordinary conversation, it was often necessary to read an excerpt several times, sometimes adding or deleting codes with each successive reading. This was particularly the case since English is not a first language for any of the ethnographers: when ethnographers were uncertain, they often included the original dialogue in siSwati. In addition, the handwriting of some was very poor. It thus often took several readings to be able to identify relevant content. The initial coding of a journal typically took four hours, but with some taking considerably longer. The coding identified 18 important themes: Several prevention themes including mention of condoms, male circumcision, concurrent and multiple partners, intergenerational sex, clinics, alcohol and stigma, as well as relationship types, motivations for engaging in sexual relationships (men and women), responsibilities of men and women in sexual relationships, and fights and arguments. Talk of witchcraft was coded as well as any insults. Coding and counting qualitative data is inevitably imperfect, and particularly so when the data are ethnographic. In informal conversations, people do not provide simple comments: they contradict themselves, they repeat their comments in different words. This raises potential biases. To control bias in the coding and counting, the coding consultant developed a protocol (see Annex 3). 2.6 Quality assurance of the data An assessment of the quality of the ethnographic data can be based on comparisons across and within journals. Reading multiple journals, it becomes evident what a normal conversation in this context is like. As already described, the excerpts that lacked credibility were not used in the analysis. During data coding, spurious conversations were identified which needed excluding from the coding and analysis. Such conversations were marked by, inter alia, highly formal modes of dialogue that are markedly different from the rest of the recorded conversations. Importantly, there were long passages of material which remained uncoded, but provided necessary context for text which got coded. The percentage of total uncoded material varied between ethnographers. However, overall, about 50% of all journal text does not carry an actual code. 2.7 Strengths and limitations of the data The ethnographic data used here provide unique perspectives on community responses to the HIV epidemic. They have corresponding limitations, some of which are similar to the disadvantages of other forms of data and some of which are particular to this method. The advantage of ethnography over other forms of data collection is that it provides information on the cultural scripts and norms that guide behaviour. These are external to individuals and expressed through informal social interactions, when people talk with each other rather than with an interviewer who is not a member of their community. In the journals, people rarely praise good behaviour; they are, however, clear about what bad behaviour is. Thus, from the criticisms of the behaviour of others, it is possible to learn what the community standards and expectations are. Surveys, focus groups and semi- structured interviews consist of questions that are set by people who are not members of the community: in ethnography, the questions, and comments, come from insiders. These questions, and the responses to them by participants in the conversations, can reveal preoccupations, beliefs, and values that would not be expressed to an outsider and thus would be unknown to those designing HIV prevention programmes. For example, as noted in the Executive Summary, there is strong support for marriage, a topic which is rarely, if ever, included on surveys related to AIDS. Moreover, a survey question requires a short, definitive answer, a yes or no, although the respondent may in fact be quite uncertain; in contrast, a participant in an informal conversation can express his or her ambivalence. 4 No form of data collection can adequately represent what people do: it is only possible to know what they say they do. This is likely to vary according to the context of the verbal interaction. Surveys and other standard forms of collecting information on responses to the epidemic are well known to be affected by social desirability bias: the respondent wants to impress the intervier.by saying what he or she believes the interviewer wants to hear. When people are talking informally with each other, there is also a presentation of self, but it is likely to be a different presentation. For example, a young man may be more likely to brag about having many partners when he is talking to his friends than when he is responding to an interviewer (or when he is talking with a church leader). The main limitations of the ethnographic reports can be summarised as follows: The ethnographers listened to conversations in their mother tongue and translated what they had heard into English when writing down the conversations in their journals. This translation step could potentially have affected the meaning of some conversations. Since the ethnographers documented the conversations after they had happened, they may have selectively recalled conversation content. Therefore, the journal texts analysed reflect the biases and interests of those who recorded them, just as survey responses may suffer from selective memory as well as from social desirability bias. Male and female ethnographers may have put different emphasis on the content or context of conversations. Contradictions and confusions characterize casual conversation (not just in Swaziland), and some discussants may have contradicted themselves or been otherwise unclear, which makes the coding of the ethnographic texts far more difficult than the coding required of a survey interviewer who must transfer the respondent’s response to a limited set of codes. The journal texts were available in PDF format to the coder, which limited the choice of qualitative analysis programmes (Acrobat was used since it allows typing on a PDF file, hence sections of the text could be highlighted, and comments could be added). In the public conversations captured in this study, it could be documented what people say they do, but not what they actually do. There were limited resources to allow the integration and reconciliation of the codes by the local coders and the international coder. The international coder noted that “It will take some time to become familiar with the way people talk about AIDS in Botswana and Swaziland, but after a while the coding will be much easier”. Nonetheless, there was considerable consistency in coding. The time frame for conducting the study was short and could not provide an exhaustive view and understanding of cultural scripts and norms of the people of Swaziland. 5 SECTION 3. RESULTS 3.1 Typology of sexual partnerships In the journals, speakers often make a distinction between categories of partnerships, evidence that the distinction is important to them. Sometimes the distinction is made explicit—“she is not my wife, only my girlfriend”— indicating different expectations of a girlfriend than a wife, and different expectations of a sex worker than a girlfriend. Even when the distinction is not explicit, the conversations show that it is taken for granted by those listening. Similarly, what the norms are is typically implicit, rather than explicit, in stories about those who did not fulfill these expectations. While there are some stories in which a faithful wife is praised, generally the stories are about those who breached the expectation for a category of relationship, for example when a wife has outside partners or a customer does not pay what he owes to a sex worker. Thus, readers of the journals, like the participants in the conversations, learn much about the norms from disapproval, rather than approval, of behaviour. Twelve categories of sexual partnerships were defined. These are listed below, along with the frequency of the comments under this type. Marriage anchors one end of the typology and commercial sex work the other. These, and intergenerational relationships - a type of partnership that elicits particularly critical comments - are discussed after a section on casual partnerships. Categories of sexual partnerships: 1. Married (couple without other partners publicly acknowledged) (26 comments) 2. Married but with a steady outside partner who is considered a girlfriend (55 comments) 3. Married but with multiple steady partners (11 comments) 4. Married but with a casual partner (76 comments) 5. Married but with multiple casual partners, married person as promiscuous (34 comments) 6. Polygamy (formal) (20 comments) 7. Polygamy but with additional multiple casual partners (3 comments) 8. Single with a steady partner (13 comments) 9. Single with multiple steady partners (75 comments) 10. Single with both a steady partner and a single casual partner (54 comments) 11. Single but with both a steady partner and multiple casual partners (talk is about having a steady partner and also sleeping around) (19 comments) 12. Single with multiple casual partners (131 comments) In interpreting the typology, it is important to note that while the categories are fixed, the occupants of the categories are not. Participants in conversations about sexual partnerships see the boundaries between types as porous: a casual relationship in which sex is regular can become a live-in relationship, which in turn has the potential to become a marriage. Correspondingly, a regular casual 6 relationship can devolve into one in which sex is infrequent and sporadic, reducing the chances of HIV transmission. Classical ethnographers do not code and count, and for good reason: ordinary speech does not fit neatly into categories. In the case of hearsay ethnography, it is even more difficult to pin down the butterflies of speech. Caution thus needs to be exercised in interpreting the frequencies above, or those elucidated anywhere else in the report. First, the frequencies are not frequencies of behaviour, but frequencies of comments about behaviour. In this study, the sexual behaviour of single people is more likely to be a topic of conversation than the affairs of married people (296 vs 214 comments, respectively; if polygamous unions are included, it is 296 vs 237). By far the largest number of conversations is about the behaviour of single people with multiple casual partners, people who could be labeled as promiscuous and are often described by the ethnographers as “sleeping around”. The second caveat is that the delineation of categories is necessarily rather arbitrary, as it is imposed by the coder based on the phrasing of the speaker, the person telling the story. The categories used in this typology are not used by speakers—no one would refer to his or her “multiple casual partners”, nor would the ethnographer use that phrase in writing the story of a man who slept around. Thus, the coder has to translate what was said in the conversation into the categories of the typology. This coding process is also necessarily imperfect because both speakers and ethnographers are imperfect. Some speakers may refer to a partnership with certain qualities as a steady partnership while others refer to a partnership with the same qualities as a casual partnership. And although the ethnographers were asked to provide quotes around the words of the speaker, some did this sometimes, others did not, and even when quotes were used, the ethnographer may not have recalled the exact phrasing. Thus, the coder had to rely on inference, albeit inference informed by reading all the journals multiple times. 3.2 Talk about casual partnerships The MOT analysis (The Government of Swaziland and NERCHA 2009) estimated that 50-65% of new infections may occur in individuals reporting in surveys to have had one sex partner only in the last 12 months. The attention of the media, HIV prevention programmes and community conversations remain focused on casual partnerships. Newspapers feature behaviour that is scandalous compared to the models promoted by HIV prevention advocates, illustrated by the following examples: • "Mzwakhe, you liar". “Eat Your Words Mzwakhe”. This is about a renowned gospel artist who impregnated a girl when she was 15. It was considered national news: there was a front page headline, an editorial, and several smaller articles. The coverage continued for several days (Swazi News. 2010 September 25, pp.1, 2 and 3). • “Miss Pigg's Peak Stuck in Love Triangle”. This story, about a beauty queen who had had two boyfriends, was also considered to deserve a feature story (Ndzimandze, Mbongiseni. 2010 July 2. Times of Swaziland: 29). The scandal of the first article violated four cultural scripts: The relationship was cross-generational, a type of relationship which comes in for much criticism in the journals; a man of prominence should be a role model; he refused to recognize paternity and provide child support; and as a religious leader, he was a hypocrite, not practicing the fidelity that religious leaders preach (in the journals, this is a common criticism of religious leaders). Three-quarters of the comments on casual partnerships are critical, and sometimes insulting. The exceptions are some young men and a few young women (see below for community comments on intergenerational relationships). Speakers are identified as young male, young female, adult male, adult female. If the age is not evident in the journal, F and M is used. “She cannot even count the number of men she's had” (young male) “Men are like dogs” (F) 7 “She is no different than a serial killer, sleeping around while HIV+” (F) “Men are the ones spreading the disease by failing to control sexual urges when they are drunk and not using condoms” (young male) “She has turned herself into a playground for boys” (adult female) “Casual sex is foolish and dangerous” (young male) “He has first-borns all over” (M) “She is food for the nation” (young male) Of the negative comments about casual sex, the most common category was “bad”: either no specific reason was given in the journal or the ethnographer did not remember it. Statements were coded as “bad” if they included words like “should” or “shouldn’t”, “too many (partners); “a lot (of partners); “prostitute” (meaning sleeping around rather than a commercial sex worker); “player.” The most frequently specified reason is that casual sex leads to HIV infection. Overheard conversations suggest that casual sex is immoral: “Sex outside marriage is wrong and will be punished by God” (adult male) “Girls getting pregnant are an embarrassment to the church” (young male) “He does not believe in sex before marriage. He is a Christian” (young male) “Young guys just sleep around without taking responsibility” [e.g. for paternity, child support] (adult male) “It is wrong to sleep with another man’s wife” (young male) Partnerships involving commercial sex workers “Prostitute” is a general term for a mercenary and promiscuous woman, and is used abundantly. It is not limited to women who explicitly sell sex, i.e. where there is no expectation that either love or a relationship is expected. Thus, the category “CSW” is based only on comments where something is said to distinguish a woman who is a sex worker from a mercenary woman seeking casual partners. Interestingly, there is very little talk about commercial sex workers—only 19 comments, the majority from male speakers. A few are positive, or at least not critical: a man who says that prostitutes are safer than girlfriends because there are no attachments, a suggestion that a rich man should buy sex rather than rape children, a man who went to a prostitute after failing to find a casual partner. Some are descriptive—a prostitute who became ill, that there’s prostitution around Christmas time, the police won’t stop the sex workers because they are customers too. The only highly negative comment is from an older woman, that a young woman is “selling her vagina instead of fat cakes”. A possible interpretation is that a CSWs occasion little comment because it is an overt market transaction and therefore neither partner is deceived. If a husband were to be discovered to have gone to a prostitute, presumably the wife would be angry (he’s risking HIV, he’s spending money that should be for the family), but perhaps this is something that she would say to him, or a best friend, in private and it would thus not be accessible to the ethnographer. Partnerships involving intergenerational sex In contrast to apparent lack of outrage about CSWs, there is a great deal of criticism of intergenerational sex. Below is an excerpt from a journal written by a middle-aged female ethnographer who had gone to Manzini and was eating at a fast food restaurant there when she overheard schoolgirls talking about 8 Sugar Daddies. The picture it presents of mercenary schoolgirls is very similar to other depictions in the literature (Leclerc-Madlala 2003; Hawkins, Mussa and Abuxahama 2005). "I went to town to my shopping, I was hungry so I decided to go to Nando's Chicken in Manizini at Bhunu Mall to buy some lunch. I bought the lunch and sat down on one of the tables. Next to my table there was a group of students from one of the schools on the outskirts of Manzini. The young girls there were five in total they were wearing school uniform. The girls were wearing blue tunics white shirts and neckties with blue and red stripes. The girls are aged about 14 to 17 years. When I sat down I greeted them and I asked them what they are doing in town at this time of the day. I asked them 'are you not supposed to be in school during this time?' One of the girls with short hair said no, Make, we are writing our final exams, we're doing Form 3.' I continued eating and I heard their conversation. One of the girls told her friends, 'Hey guys, I have to leave now, my 'Sugar Daddy' imphunga said we should meet today so I should leave early.' She needs to call him so that they can meet early so that she can maintain her curfew at home which is 5:30 p.m. She said her mum took her cell phone last week and she kept it and told her she will get the phone back when she has finished her education first, because the phone makes her crazy. The girl told her friends that her mother is playing, because her Sugar Daddy promised to buy a new blackberry phone today. One of them, who seemed to be the youngest, said 'My friend, how will you feel when you are married and a young schoolgirl will snatch your husband away from you?' The other girls laughed at her and said, 'You're a stupid girl, you don't know that men were meant to be shared, indvodza lichatsa iyaluyiswana (meaning men are like meat, you buy it and leave for other women). Another girl said the reason why this younger girl talk like this is because she has not started having the forbidden fruit (sex). she said sex is good and healthy and it makes one glowing. The youngest girl said it is better because she is safe, she is saving herself for her husband. She said her mother tell her that she should not have sex at early age. They asked her what age did her mother tell her she should start having sex. The youngest girl said her mother told her that the right time is when one is married. One girl told her that your mother hates you, she wants to have the good things alone and at what age did our mother get pregnant and was she married? The youngest girl said it was none of their business. The girl who said she wanted to meet her Senator today said 'It is the last day we are buying lunch. Every time we pay for you and you never have money. It means you are sucking us.' She said 'You should go to your mother and let her buy for you. Do you mean you will tell your mother when you have a boyfriend?' The youngest girl stood up and tell her friends that she was begging for their food and friendship, and she left. The rest of the girls laughed at her and said 'go mama's girl'. The rest of the girls then started planning on what they would be wearing during the Simunye fun fair and what kind of booze [they] will be drinking. The girls then started talking about their boyfriends. They said they don't want to go with their boyfriend, they want to get new ones at Simunye. One of the girls said she wants to show herself at DJ "Black Coffee" on Saturday and [the] media will take her photo. The two girls left and go home. The conversation ended like that. I finished eating and I have to leave". Had the ethnographer told this story to her friends, it is quite likely that they would have been very critical of these young women. A review of prevention programmes in Swaziland includes the statement that “social norms are currently permissive of such [intergenerational] relationships” (The Government of the Kingdom of Swaziland and NERCHA, 2009, p. v). The journals contradict this: the vast majority of comments in conversations about intergenerational sex are highly critical, as shown in the figure below. 9 Figure 1: Attitudes towards intergenerational sex 80 70 60 50 for (N=13) 40 against (N=75) 30 20 10 0 Women, and especially adult women, are more critical about intergenerational sex compared to men, as shown in the next figure. Figure 2: Attitudes towards intergenerational sex by gender and age 40 35 young females: against (20) 30 adult females: against (37) young males: against (7) 25 adult males: against (10) 20 - young females: for (8) 15 adult females: for (0) 10 young males: for (2) 5 adult males: for (2) 0 The criticisms are directed equally at young women with older men, and young men with older women, and are often quite harsh. The frequencies of negative and positive comments about intergenerational sex are shown in the next graph. 10 Figure 3: Attitudes against intergenerational sex by types of partners 25 20 fy+mo neg 22 15 mo+fy neg 22 10 fo+my neg 9 5 my+fo neg 8 mo+fy pos 6 0 Negative and positive attitudes toward intergenerational fy + mo pos 5 sex. F-female M-male Y-young(er) O-Old(er). Order of my+fo pos 2 entries reflects power balance. Eg. Fy+mo Neg means negative attitude towards young women who sleep with older men. Mo+fy neg is vice versa “People who sleep with school children should be killed” (adult F) “This girl is in love with sugar daddies, so sex with her is dangerous” (young female) “He should propose his age-mates” (young female) “He has money, why doesn’t he get girls his own age?” (young male) “What is wrong with you, sleeping with old ladies from the shebeen?” (adult male) “She sleeps with boys as young as 25 years, how can you enjoy sex with someone your child’s age? “ (adult female) “He is not supposed to have young girlfriends” (young male) “If men would stop chasing girls and stick to their wives the country would survive” (middle-aged woman) Much of the criticism has a moral tone, suggesting that disapproval of age-discrepant relationships may have been seen as contravening the norms of sexual behaviour in the past as well. 3.3 Motivations for partnerships The most frequent comments about motivations for partnerships mentioned love, money, revenge and sex. Although in principle these motivations are distinct, in practice they are tightly connected, and thus difficult to code. For example, a young woman said she “would only have sex with a man who she loves and who could provide for a baby”; a young man said “love doesn’t exist: we say we love to have sex”. There are more descriptions of the first stage of a partnership in men’s journals than in women’s. In these descriptions, it seems that when a man approaches a woman, he is very likely to begin by saying “I love you” which is likely to be understood by the woman as “I am interested in having sex with you”. They both take it for granted that at some point resources will be provided by the man to the woman. 11 Gender influences the attribution of motivations for partnerships. A debate among young men and women provides a partial illustration: Young male: “It's a normal Wednesday, boring as usual at around 1:45 p.m. it is me and other friends of mine we had been together for the better part of that day we had nothing to do all that day, this is due to the fact that we are all unemployed thus we usually spend the day telling stories and other related matters of life. While we eat together we end up getting a few drinks from nearby shops at………. Sometimes we have our female friends; this makes our conversations become more interesting as we get to know some views from the female part”. We were busy arguing about who gets jobs easily, it is the male or female gender? One female friend said ‘It depends on the type of job if it be the office jobs or one which needs muscle.’ One male friend of mine said 'You females are considered first, it doesn't matter on the nature of the job.' The majority of us including some of the girls agreed to that, this was evidenced by us nodding our heads or by word of mouth. One male friend said 'Let me put it to you as an example. Let us assume that we applied for the same post with me having high qualifications than you (referring to the lady with the different view to us). Do you know that you will be the first to be accepted over me.' She argued, 'What's the cause of that?' One other lady jumped up to answer before anyone interrupted and said 'We ladies are always above guys, it has always been the fact that once we are offered the jobs we end up having affairs with our superiors. So what will they be looking for among guys?' Said one friend of mine, who we know always talks some nonsense some times, That means at all times you end up having intimate relationships with a family man and spread very dangerous diseases.' Said one female friend 'By which you mean to tell us that the diseases are carried only by females? Even the males, I can boldly say, you are responsible for spreading the virus, because everyone the boss hires, it is obvious they will engage in illicit sexual acts with a newly hired female. The one responsible for the spread of the virus becomes the man, regardless of where he got it from.' I received a call from a girlfriend of mine who wanted to see me immediately. I had to leave them there with the topic on its climax, with one female friend up on her feet due to the state of the topic. I went to my girlfriend and we chatted for about 15 minutes after which I went back to my friends. Upon arrival I was slapped with a question on my face before I even made myself comfortable. Said one lady, aged in her early 20s, positing the question to me, 'If you could be honest enough to yourself, who would you hire as your first option between male and female, given the chance to be a boss in a certain institution?' Before I answered I couldn't help but laugh. One lady said 'you have just made it clear by your reaction that you would go for the ladies'. Said one male friend, 'Honestly, how will I go for a male over a female? To a female it is more like an investment, one way or the other.' 'Exactly! said one lady. (Meaning 'Now you are of my idea). Everyone, honestly, will go for the ladies unless one is possessed by what is nowadays the demon of bisexuality.' One other lady just jumped to comment on that and said 'Excuse me, who said that is a demon? This is something you can't just decide overnight, it is something one comes with from his or her mother's womb.' ' No, bafeto, let us not fool each other. This is just a fact. We are living in the devil's era. How can you have sexual feelings for one of your sex?' one male friend said. Another female friend said 'As I mentioned earlier, this is something one comes with from birth, it is not like one morning you just wake up and decide you are going to be a lesbian or a gay.' Said one other guy amongst us 'Hey stop making fun of us. We have heard before how a mother of two gets hurt by her fiance that she resorts to gayism.' Said one other lady who had been not so vocal, 'It seems like that is the best way [rather] than being fooled by a guy , and then again gays are honest with their partners.' Said one other female 'Friend, don't fool yourself, because even if this demon can be 12 spread like wildfire, there will also be cheating amongst couples. That will also mean the spread of diseases.' Said one male friend with a worried face, it has a writing [it was written on his face], 'This aids demon seems to be difficult. This thing won't be easily controlled. Where is there a place one wouldn't resist? Even if one uses a condom to his or her partner, that won't last, because one would end up having rust with his or her partner. That partner is maybe having a secret affair with someone [and] they are not using protection. By the time there is this chain, that means the spread of the virus.' Another male friend said to another friend of mine 'Yah', mockingly, meaning you've just put it straight to the point. 'I wonder that what you preach is what you practice. I believe you are honest to your partner, please don't infect her with this deadly disease she is just an innocent somebody.' In response to that, one said 'It is no use to lecture me only because we don't know what she is doing in her secret places. It is worst to those who seem innocent because they are the ones who are most dangerous. ' One ladyfriend stood up and went straight to the shop. To our surprise she bought a liter of coca cola. This was due to the fact that she hardly [ever] spent her money, she always complained of being broke, even if she had a lot of cash. ….. As one of my male friends said, 'Hey guys, I won't lie to you, I can never stick to one girlfriend. They too have other secret lovers. You will end up committing suicide on being discovered that she has been cheating you.' [Another man said] 'I do agree with you on that, as I have been a victim of a love triangle, not even just once, but that doesn't mean we have to resort to having many lovers in our lives, that is the reason the spread of this deadly disease is in its best (so great). It is a pity we are not the same as we cannot do the same thing [we are all different]. If I were that much superior I would be glad if we all be honest to our partners.' We were all surprised to hear such from somebody of his caliber. We didn't expect to hear such words of wisdom from him since we knew his history with girls. Said one lady friend who had been so active in the topic 'I am so surprised how you have changed from worse to better, you have been into girls in the past, changing girls in even just one day, I wonder if you used that protection?' He replied 'My friends, if you're careless about your life we will soon bury you. There are too many things to live for. Fortunately I got rid of my wrong doings early enough. I got tested and luckily I'm HIV negative. ' Said one other male friend, 'I will never ever do that.' One female friend said 'why would you do that [never test]?' He simply responded with confidence, 'There is one problem of testing. Lt's assume I just wake up tomorrow and decide to go for an HIV test. After which I realize that I'm HIV positive. Is anyone aware that the minute you discover you are positive that's the moment you fall sick? Now I don't know my status but I'm fine and feel healthy. ' Said one other male friend, 'On that account I can agree with you. Really, by the time you discover that you are positive, the mentality changes completely, and you consider yourself dead.' Seemingly the females were all shocked at this revelation. One female friend who was so much shocked said 'One foolish thing about boys is that they seem to care less about AIDs. Again they are all the same. If you are about to engage in sex with a guy usually its us girls who bring the idea of a condom. If you don't tell a guy to use a condom, he won't bother to use it. He doesn't care who you slept with before or your HIV status.' The topic turned into an argument. The females defended themselves while the males also supported their part. The guys didn't want to admit they were responsible for the spread of HIV while the girls put the blame onto the guys. Said one female friend, 'It is for real that you guys are responsible for the spread of Aids.' The guys all cried foul to that as I laughed out loud. One male friend said 'I disagree with what you have just said. You females are the ones responsible for the spread of Aids. If you were not into sugar daddies, married men, this situation would have been better. You are easily tempted by money and other valuable things offered to you and you end up falling for too many men.' Said one other female friend, 'What would you do if one offers you a lot of money and suddenly he gives it to you, I mean hard cash, money compared to my parent's monthly salary put all together? On top of that he buys you everything you've always dreamt of? One other male friend of mine said, 'Then you agree with me that you are responsible for the spread of diseases? Why do you let someone manipulate you? That 13 means he doesn't love you, his mission is to get what he wants and then he leaves you for others. You will end up waiting for another man on the lookout because you are used to getting paid without working. Then the ethnographer gets a phone call, as he leaves he hears a male friend saying 'Hey stop it, just agree that you are responsible for the spread of Aids, just get rid of these married men.' Women’s motivations The code for what motivates women to have partners covers both motivations for marriage and motivations for a casual partnership, and includes individuals speaking about their own motivations and perceptions of the motivations of others. Sometimes the speaker refers to his or her perception of the motivation of an individual, but far more common are comments about the perceptions of women in general, as in “women just want money from a partner.” Money: Accusations of girlfriends having Sugar Daddies; young boys can’t afford to treat girlfriends; friend talks about using her boyfriend to get money for a new phone and clothes; people [girls] will leave you if you don’t have money; women will abuse you if you don’t have money and you can’t get a girlfriend without money; woman admits to turning to prostitution for economic survival; have to look good and throw money around to get girls; girl can’t have sex for nothing; mercenary young schoolgirls; girl prefers to wear old clothes rather than have sex for money; people think she is stupid to break up with someone who gives her money; there is no love these days, everyone is looking for money. Sex: Good sex will keep a woman from looking around; they are hungry for sex; because they are not sexually satisfied; young wife will leave you when you get diabetes and can’t perform; girls will leave you if you can’t provide sex; sex-starved in polygamy; best to just have men over for sex and then send him on his way. Love: She’s not in love with the rich guy and leaves home to live with the poor guy; loves him, is afraid she’ll lose him if he knows her status; sugar moms want sex and affection; There were a total of 278 comments on women’s motivations for partnerships. Sex accounted for about 28% of the comments and love accounted for 14%. The next largest category, 7%, was about women who chose a partner in order to deliberately pass on her infection. The rest were about money either as the single motivation mentioned, or a combination (sex and money, sex and love, love and money). Men’s motivations The categories of men’s motivations for having partnerships are the same as women’s, although the order of frequency of comments differs. In addition, there are more unique motivations for men’s partnerships. Sex: Driven by demands of the flesh; impossibility of abstaining; “you know how we men are”; on a man taking a partner because wife doesn’t satisfy him sexually; good sex is a stupid reason to get married; real men need multiple partners, it’s just so; “it’s useless being in a relationship and not having sex.” This motivation is by far the most frequently mentioned in the conversations (66 comments where specific motivations are given). Money: Boys these days are looking for someone with prospects; Sugar Moms are easy; thinks men are after widows to get money from the dead husband’s estate; widows common targets for men; gets new partner just for her money; she buys him drinks (33 comments). Love: “in love with a girl”; love at first sight; “he’s in love”; doesn’t love her anymore; loves his wife (24 comments). 14 A small number of comments attributed men’s motivation for a partnership to be care (earlier, it was stated that a major responsibility of married women is to care for the husband). Of these “care” comments, most were men looking for a wife to care for them or a girlfriend to care for him when he works away from home or is sick. There were 11 comments about men motivated to seek a partner in order to spread the virus . Comparing women’s and men’s motivations as they are perceived by their fellows, equal number of comments present men’s motivation to be sexual, and the same goes for money. Interestingly, no women are said to believe that either men in general or a specific man in particular have partnerships because they are in love: love as a motivation is attributed to males only by males. Both men and women are said to want children, but slightly more men than women talk about this as a motivation for a partnership. Only men talk about having women to prove their worth in status (12 comments) and about competition/conquest as a motivation (11). 3.4 Dynamics of sexual behaviour Compared to other countries in Sub-Saharan Africa, Swaziland, as well as Botswana and South Africa, have a distinctive pattern of marriage and a high frequency of childbearing outside of marriage (van de Walle 1993; Hunter 2010). Margo Russell’s (1995) analysis of the data from the Swaziland Census of 1985 shows that only half of the women age 20 and older were married, most of them traditionally rather than in a civil marriage. Of the unmarried Swazi women age 20 and above, a third of the women had children; in one village in which she conducted individual interviews, half the women had children from their husband at the time of marriage, and 41% had children from other men at the time of marriage. Thus, both the census data and the interviews suggest that non-marital partnerships were not uncommon. Russell draws on ethnographies by anthropologists done in the 1930s to propose that marriage was a “far weightier thing than procreation” in the past, and that this pattern goes back to the pre-colonial period. Russell’s explains the decline in marriage and increase in out-of-wedlock childbearing by noting a decline in the willingness of men to pay lobola so that they could get their children, positing that this was associated with a decline in the value of children. The most thorough examination on sexual dynamics in southern Africa is that of Mark Hunter, a scholar of South Africa. Writing on similar patterns of non-marriage and out-of-wedlock childbearing there, he offers more detail and a more compelling explanation. Evidence from court cases and oral testimonies collected at the turn of the 20th century, and from the work of anthropologists of the 1930s, indicate that even before the period of migrant labor there was “a certain level of acceptance around women having more than one soma1 partner, although it is true that those overstepping the mark could be chastised as being izifebe (pl. of isifebe, a loose woman”) (Hunter 2005). It also appears that at this time the “attitude of unmarried women having multiple-boyfriends” was “’the more skulls the better’” (Hunter 2005, citing Wilson 1936). He attributes more recent changes in sexual behaviour associated with the AIDS epidemic to the steep decline in employment opportunities for men which, he finds, makes marriage more concentrated in the middle class, since poorer men do not have enough to pay lobola and support households. Correspondingly, the lower marriage rates make women more dependent on the men who do have money (Hunter 2007). Hunter’s emphasis on employment is consistent with a detailed and careful analysis of Alan Whiteside (Whiteside et al 2003). 1 Partner with whom sexual intercourse is held 15 3.5 Norms about relationships Here, the relative importance of various responsibilities that a man or a woman is normatively expected to fulfil is discussed. Again, criticisms outnumber praise. This does not indicate that breaches of obligations are more frequent than good behaviour, however. Just as newspapers and focus group discussions feature scandalous acts, so also it would be far more interesting to talk about someone who misbehaved than to talk about someone who fulfilled her responsibilities. First, women’s and men’s responsibilities were grouped in terms of the categories that occasioned the most comments, and listed by frequency. The comments were not interpreted as equivalent to actual behaviour: it would not be appropriate to count, for example, the number of men who claim to have many partners or who are said by others to have many partners. Rather, the frequency of comments is counted, since it is these that contribute to cultural scripts, not whether the story about an individual’s behaviour is true—which, in any case, is not known. Some of the comments were summarised to provide brief illustrations of what is included in a category. Since all of the excerpts come from conversations about AIDS, the comments of participants give an indication of the range of topics that were associated with AIDS in the perceptions of the speakers and/or the ethnographer. Married women Children: Bear children, care for them, raise them well (women are often criticized in absentia for putting their own needs above their children’s). Husband: Care for all of husband’s needs, be supportive, tend house, care for him when he is dying of HIV. Husband: Be submissive; Obey husband; follow his rules and regulations; do not refuse him sex; do not ask where he is coming from when he returns late at night, be patient with his infidelities. Husband: Respect husband. Be faithful; do not embarrass husband by adultery; widows should not take another man to the deceased husband’s homestead; wives should dress properly, not wear trousers. Married men Children: Shirking responsibility to children; irresponsible to have so many kids and not take care of them; failure to care for a child that leads to the involvement of the police; not taking care of children’s needs. Mothers of the children: Men do not think of what will happen to the wives after they die (multiple comments); blamed for infecting their wife with HIV (that a child may be infected by the husband through the mother is never mentioned); ‘he will regret dumping me and his child’. Wives: Man’s responsibility to satisfy his wife sexually; a man has to satisfy all his wives. Model Man: ‘Having HIV has made him more responsible. A real Man;’ being with kids and teaching them; man takes offence at suggestion that he is not looking after his wives and children; even men have to behave themselves because of AIDS. There are more comments about married men’s responsibilities than about married women’s responsibilities; again, this may be due to the greater number of male than female ethnographers. Most of the comments about men are criticisms, whereas for the married women a larger proportion was positive exhortations reflecting ideals: be faithful, be submissive, be respectful. 16 A key finding of this research is that the norms regarding marriage, either civil or traditional, are the template other types of partnerships: quasi-marriage (informal partnerships where the man and woman are living together), casual partnerships (from steady relationships to commercial sex) are diluted versions of marriage. A pictorial representation would show a married couple in bold, with more casual partnerships becoming progressively faded as they go from quasi-marriage to frankly commercial sexual relationships. The implication is that to understand casual partnerships of whatever type, it is useful to bear in mind that the heterogeneous category of casual partnerships is not a separate genre of relationships; rather, the relationships are on a continuum stretching from legal marriage to frankly commercial sex. While it is usually clear whether the relationship that is the topic of the conversation is casual, it is often unclear whether people referred to as a “husband” or “wife” are in a marriage or a less formal quasi-marriage. The distinction is apparently more important for the female journalists than the male, since they are more likely to note whether or not the speakers or the person who is the topic of conversation is married formally or not. The strength of the norms, and thus the degree of censure when they are broken, varies by the type of relationship, the particular norm, and the gender of the offender. Thus, for example, sanctions are more severe for those who are formally married; a man who does not support his children is blamed more than a woman who does not fulfill her domestic obligations; and an unfaithful woman comes in for more censure than an unfaithful man. An excerpt from the study data describes the public reaction when a man did not provide support for his child. The ethnographer is male, he goes to a bank and happens to witness a confrontation over a man’s obligation to support his child, a norm about which, shown later, there is little disagreement in the study communities. Parenthesis are comments made by the ethnographer, brackets are clarifications by the analysts. Here, as elsewhere, the names of places are deleted and the names of individuals are changed. It was around half past ten in the morning. I was queuing at the ____lst National Bank. There were two people in front of me in the queue, the first one in a brown jacket [he describes her clothing], the second also a lady who was tall, wearing a long black and red skirt. [he describes her clothing]. On her back she was carrying a baby between 2 and 3 years. As the lst lady in the brown jacket moved out, the tall lady, (the 2nd one), moved towards the counter and started loosening the baby on her back. She held the baby on her arm and looked at the guy (teller) who was sitting on an office chair, wearing a white shirt, and a gold watch on his left wrist, and a gold ring on his second left finger. His dark face smiled at the lady who was carrying the baby. The young teller spoke first: 'What can I do for you my sister? You want to open an account or you want to deposit money?' The lady carrying the baby replied in a low tone, ' Caleb, don't ask me that.' In a very soft voice, Caleb said 'What must I ask you? You want to tell me what? I want to continue with my work sister, so tell me what you want help with, you are delaying me.' The lady responded in a loud voice, 'Caleb, do you know how much this baby had delayed me? Please don't fool yourself, brother, it's above my shoulders, you are working here at the bank, smiling here at everyone in front of you, but your baby is struggling as if you are dead. I'm suffering the consequences of being a parent alone, yet you are alive. Now you ask me if I want to open an account to deposit money---am I new to you Caleb?' In a small voice. Caleb said to the girl, 'My sister, as for now I am at work, please don't delay the customers behind you.' The lady shouted as if something was pressing her. 'Caleb, I don't care about your customers, what does this child eat and wear? Why should I suffer alone when you are working and handsome to all ladies in town? You are a fool, all men struggle for their children, but you alone struggle for useless things, you go around dating different ladies in this place, your child is dying because of hunger. Nx!' In a calm voice Caleb said, 'Kiki, please listen, listen Kiki.' Thembi shouted louder, fastening her baby on her back. 'Caleb, how can you tell me to listen when both my ears are widely open.' Caleb stood up from the office chair, opened and closed drawers in his office cubicle, taking and putting a red pen. 'Kiki, 17 please wait for me outside. As for now I'm at work. We will sort it at lunch' said Caleb with his eyes widely opened and looking to see if we (customers) were looking at them.' 'Forget it', she said, clapping her hands, 'I'm going nowhere. For two years you don't know what your child eats and wears. I have to go around looking for food for your child. Like it or not you will support your child and me. You know very well that I'm an orphan. The best thing I will do now is to go to your manager's office and leave your child, you'll find it there.' 'ENOUGH IS ENOUGH' shouted Caleb, moving towards the door which was written manager's office. Caleb shouted, Thembi Thembi please don't go there, come back, I will give you the money you want.' Thembi didn't say anything to Caleb. She fasted her baby and she raised her hand to the door handle on the manager's office and entered inside the office. In a vibrating voice, even shy [ashamed] to look at the customers, 'I'm sorry for delaying you, please wait a little bit, someone will be helping you, ' said Caleb, going to the manager's office. People were laughing and clapping hands. A short dark guy said "this is not a bar but it's a bank, please lower your voices, you are disturbing the others (meaning the bank workers). Behind me a woman around 50 years said in a low voice 'This girl has done a great job. Once man impregnates us simply [he] forget about us. This lady is saying she is an orphan, that is such a painful thing, what are they eating if this 'clever' (meaning Caleb) is not supporting them, especially the child.' Soon another teller entered into Caleb's cubicle and he was wearing clothes [like Caleb's, except for the watch and ring. I moved towards the counter, people were talking in very low tones. As I moved towards the door I heard a very deep voice saying 'I salute woman, you can lose your job because of them.' The old woman in a polite voice said 'things have changed, we have rights, the lady is fighting for the right of the baby, men are cowards.' The bank was full of laughter and I moved out of the bank. This particular ethnographer writes better than most, but is similar to others in his ability to present a vivid picture of a situation that is, the analyses below show, not uncommon. A young man with money (the salaried job, the gold watch) who is married (the gold ring on the 4th finger of the left hand) is confronted by a young woman. She claims he is the father of her baby, a claim that he does not refute, at least not in this interaction; he may, however, deny paternity later if she were to sue for maintenance, which both parties would know is a possibility. Although it is shown later that women are normatively expected to be submissive, the two women who speak in this excerpt--the young woman herself and the older woman who supports her (and makes comments on the behaviour of men in general)—are certainly not submissive. It is likely that men waiting in line who have a girlfriend with a baby, or a girlfriend who might become pregnant in the future, will remember this incident, and repeat it in detail to their friends—the public shaming, especially since it was done by two women, who are expected normatively to be submissive but clearly were not. The incident is arguably more powerful than a media depiction of such a situation would be because the participants can identify more strongly with people like themselves than with actors whom the audience knows are playing a role. The excerpt above also illustrates the advantage of a method that captures what is said when people are talking on their own, without guidance from an outsider. In other approaches, an outsider, usually more educated than participants, is directing the flow of the interaction, asking questions and cutting off topics that deviate from her interview guide. Thus, what is not available to the analyst is what is not said, which may be as important as what is said. For example, in this excerpt no explicit mention is made of AIDS, but the association between casual partnerships and HIV transmission is likely to have been made by all present. There is widespread consensus that men have responsibility to support their children. The most frequently mentioned type of support is material, but there are occasional mentions of the responsibility to set a good example in terms of sexual behaviour or guiding the children as they grow up. When a man is criticized for failing to support his children, it is in terms that provide supportive arguments for the HIV prevention messages. Most commonly, the failure of a man to support his children was attributed to 18 his having other partners, as well as his having other children from other partners; in such cases, whatever resources he had would be viewed as diverted to the new partners and the new children, and the new mothers. Alternatively, the man may have continued supporting children from a marriage but failed to support children from non-marital relationship. Married men are criticized for having died without making arrangements for his children’s future support; since participants in a conversation would recognize the difficulties of a poor man who earns very little would have in providing future support, it is likely that the criticism is of the failure of the man to avoid death—he knew how to prevent HIV infection but did not do so. Married men also have a responsibility to their wives, but this is most likely to be as mothers of their children. There are justifications for running away from a wife if she behaved badly (didn’t obey him, refused him sex and particularly if she was unfaithful), but he cannot run from his innocent children. Because few wives are likely to be unable to bear children, the issue of what a man’s responsibilities are to a wife who is not the mother of his child simply does not come up. A man’s responsibility for his parents receives surprisingly few mentions. Perhaps this is due to the low and uncertain income of men in the sites in which the ethnographic journals were written, such that few would blame him for not supporting his parents. 3.6 Talk of marriage Talk of marriage is less frequent than talk of casual partners. It may be that talk of marriage is more likely to be private, or that marriage generates less scandal than casual promiscuity. It needs to be noted that in censuses and surveys, it is often not clear whether a respondent who reports she/he is married is referring to a formal marriage (civil or traditional) or to a consensual union. In the journals, the distinction is even less clear. A speaker may use the term “marriage” to refer not only to formal marriage and consensual unions, but also to a steady partner, or even a very casual partner, as in a conversation about a drunken man who proposed marriage to a girl on the street. Generally, however, it is clear that “marriage” refers to either formal marriage or quasi-marriage, a relationship where the couple lives together and has children together. The comments about marriage were coded conservatively. The negative comments were not absolute, and some can be interpreted as positive towards marriage. For example, if a speaker says that someone is too young to marry, it was coded as “negative”; similarly, comments about marriage being difficult were also coded as negative. Cheating, as well as anxiety about the possibility of cheating in the future, are the topic of much talk, but the difficulties of marriage also concern the costs of weddings. Despite the conservative coding, and in contrast to the largely critical talk about casual sex, most of the comments about marriage were positive. 19 Figure 4: Frequencies of positive and negative comments about marriage 160 140 120 100 positive about marriage (N=147) 80 60 negative about marriage (N=75) 40 20 0 In all the conversations about marriage, there were a small number of generalizations to the effect that remaining where remaining single was perceived to be preferable to marriage. When the story was about a specific person, fewer than one percent of the comments justified not marrying. A third (50 out of 147) of positive comments about marriage are unspecific: that is, no specific reason was given for viewing marriage favorably. It may be that just as there is a common understanding of why casual affairs are “bad”, there is also a common understanding of the advantages of marriage. The frequency of various arguments in favor of marriage is shown below. Why is marriage viewed so positively in a society where so many people in their 20s and 30s are not married? Notably, the largest specific category of positive comments are about polygamy; not surprisingly, most of these are made by men. Several of the reasons for marrying suggest that it is aspirational, e.g. the speaker says that he or she wants the love and trust of marriage, to make her parents happy, to have a white wedding, “I want to choose my own wife”. Other comments suggest more urgency, e.g. “I have children already and I want to marry the father of the child”; “the time for her getting married is getting closer”; that the person does not approve of sex outside of marriage and is thus waiting for, and presumably expecting, marriage. The social sanctions against marital infidelity are buttressed by legal sanctions, whereas there are no legal sanctions for casual partnership or frankly commercial sex. For example, while cultural scripts are available to criticize men who do not support their children, regardless of the type of relationship that produced the child, the sanctions (and the criticisms) are less severe for non-support of a child born from a casual relationship than for children born in a legal marriage or from a couple living together in a quasi-marriage. For these reasons, marriage would be more attractive to women than to men: a woman who says “I want to have children in wedlock” may be taking for granted that there is recourse, either social –such as appealing to his family--or legal, if the father does not support the child. Yet young men and young women are equally positive about marriage. 20 Figure 5: Attitudes towards marriage by gender and age 60 young females: pos (49) 50 adult females: pos (21) 40 young males: pos (47) adult males: pos (40) 30 young females: neg (25) 20 adult females: neg (28) young males: neg (13) 10 adult males: neg (8) 0 Those who are least positive about marriage are adult females, yet in the sites where the ethnography was conducted, married women distinguish themselves from other women by covering their head. This suggests they are proud to be married and want others to recognize their status. The major sources of their negativity are infidelity, or potential infidelity, and HIV risk, which in community perceptions are intertwined. Their most frequent complaints are about polygamy (which can be considered as infidelity) and HIV risk in marriage, also perceived to be associated with infidelity. 3.7 The 4 C’s: Circumcision, Concurrency, Condoms and Clinics What people say to each other about the themes of HIV prevention campaigns is particularly important in a setting where there is considerable skepticism of authority. A clinic nurse might promote circumcision, or a billboard warns against concurrency, but these sources are often considered to be hypocritical: they say one thing but do another. Below is an excerpt from a long and meandering conversation between two men, friends, on a bus. In a context where there is skepticism about the motives of those in authority, the importance of stories like this in influencing change is, arguably, as great as or even greater than prevention messages disseminated in other ways. The excerpt below illustrates this skepticism. Two men are talking on a bus. The ethnographer writes that they seem to be friends, in their 40s, and by their clothes they are coming from work. One has a small bottle of alcohol, and they talk about this for a while. Then they ask the driver to turn on the radio, and they listen to the announcer. It is the last day of November, just before AIDS Day on December 1. After listening, they interpret what they heard: The newsreader said tomorrow at 2000 hrs we are asked to switch off all lights so we can imagine how is the world going to look like when we do not take care of ourselves and stand up and fight HIV and AIDS.' The man with the black cap responded, 'The Aids fighting people are insane, now I do believe that Aids makes you to come mentally disturbed. They know where this disease comes from. Why do they not go to the whites and ask the ones who came for[with?] it so that they can get the cure for it and pay him money. Why do they combine with us [tell us?] so that we switch off our lights? They want to rape our children. And they want us to cry that our chicken was stolen during that hour.... as they are broadcasting this even the culprits are planning a way to start their robbery. As for me, tomorrow will be my last night to spend with my wife. How is she going to see me [without light] and enjoy being with me? In that lone night I will be sober like a judge.' 21 The man with the yellow cap said, 'Now it is each and everyone's responsibility to hold on to what he have. Everybody has to tie his or her dogs. Nothing can be done, only God knows how is this disease going to be removed from this earth. If we switch off lights the disease is spreading at a high speed, because people will be seeing an opportunity to sleep on the side of the roads because of the darkness of that moment. As they say, they will be having a meeting. In that meeting some of the delegates there will have come with their own motive. Others say it is a chance of being with my friend from that far country, at long last to see her [from] the previous year.' The man in the black cap said ‘they train people each and every day to teach about AIDS, and some have formed groups, are fighting AIDs. But at the end, the very same people in the group, they spread the disease because as they practice to teach safe sex, they sleep around with each other without considering what they teach in the community. They get skills on how to spread the virus more to each other.’ The man with the yellow cap said ‘Yah, ushaya khona (you are true). These people introduced condoms but they were the first ones to jump them and sleep without them. The people from the cabinet, you hear that they are found sleeping or having affairs with other men's wives and impregnate school going girls. And the other side, they tell us to abstain and use condoms.’ The man with the black cap said, ‘you remind me of something. I am a candidate [for spreading HIV]. I was once a youth leader that was teaching about AIDS, but we were not practicing what we were teaching. When we were doing the plays, the one whom I was playing with we ended up having a bond and obviously were sleeping still doing what we were telling the people not to do. It was happening mostly when we have camps and doing the practice of what we were teaching the community. One thing you have to know with this groups, everyone has got his or her boyfriend. They laughed. The conversation became serious in a way that everyone on the bus were taken by what these two friends were talking. You hear people saying ‘they are telling the truth’. A certain woman failed to hold herself, she said ‘From today my child will no more go for the youth groups she always tell me about.’ Everyone laughed on the bus. The woman was angry. The man with the black cap said ‘Now everybody has got the information. I do not think there are people who still do not know about the pandemic, but we still make girls pregnant and continue dating other men’s wives. I cannot blame it, it is our pride to have more than one partner. Now a man in the black hat said ‘everyone has the information about AIDS.’ The man with the yellow cap said it’s just that these people are trying to stop water that is already flowing over the dam, instead of diverting the water from where the stream starts. Once you try to start that water it will try to form another dam. By which I mean they teach about the pandemic just here around town, they do not consider the rural areas. Then the people from the town go to the rural areas and spread the disease, because those people know nothing about the pandemic. If you go to the rural areas a lot of young girls are having kids. If you ask about their fathers you will hear that he is at Matsapha or Manzini. The man with the black cap said ‘I’m tired of these people. I’m tired of these people. Everywhere you go is now Aids. They have taken the MTN slogan to advertise aids. People are forming anti-aids clubs. It is like now people are making money through AIDs. A lot of organizations have been formed just because of Aids. Do you know what the Mozambican said?’ ‘No my friend‘ said the man in the yellow cap. He [the man in the black cap] said God promised to take us one by one but now he is taken us in trays and dozens, then you see that we are all created to die, not even one of us will die on the grass like a grasshopper. They must let the dying type die and leave us who are strong to continue to enjoy life. The conversation lasted for about 1.5 hours, almost the whole journey from Manzini to Simunye. Unfortunately, the man with the black cap alighted before we reached Simunye. Everyone was left murmuring in the bus up until we reached Simunye. Male circumcision Following the reports from three randomized controlled trials in Sub-Saharan Africa of the protective effects of male circumcision, a pilot study was conducted by The Family Life Association of Swaziland. The study surveyed men in Mbabane, asking them about their knowledge, perceptions and practices. In 22 addition, a Male Circumcision Task Force was formed in Swaziland. The Task Force organized events to promote the practice, including what came to be known as “Circumcision Saturdays”; in addition, the Family Life Association of Swaziland collaborated with a non-governmental organization, the Jerusalem AIDS Project, to bring Israeli doctors to Swaziland to train local doctors (Grund 2010). In 2007 the Government of Swaziland published a national Male Circumcision Policy that outlined a circumcision campaign directed at men 15-24 as well as newborns. This was followed in 2009 by a five-year plan to generate demand for male circumcision, with the objective of circumcising 80% of men in this age group. The study in Mbabane showed that 94% had already heard of MC, primarily from official sources, the media (45%) and friends and relatives (37%). Few, however, had been circumcised (15%). When asked whether MC reduces the risk of HIV, 53% said no; when asked if it improved sexual satisfaction, 84% said no. This increased when the question was phrased in the negative. Then, 96% said it did not improve sexual satisfaction. In addition, 51% said the process was painful and 18% said healing would take a long time. By 2009, only 8% of a national sample of adult males had been circumcised (NERCHA 2009:vii), considerably lower than the percentage of men surveyed in Mbabane. By 2010, when the hearsay journal study was conducted, those who talk in the journals about the circumcision campaign perceive it to be “all over”, “in the radio and in the streets”. In only one of the many excerpts on circumcision did it appear that a young man might have been unaware of the campaign, and he was mocked by the others. Many of the conversations described the experience of a friend who had undergone circumcision. These personal stories, both positive and negative, are likely to be particularly influential. The following excerpt illustrates how prevention topics arise in settings far from the formality of a clinic or a life-skills training, where an authority is disseminating a message. Here, it is in the changing room after a soccer game. The ethnographer’s team, which lost the game, is in the changing room with the other team. The coach, who is from South Africa, is present, and berates his team for losing the game against a weaker team. Since the men are washing and dressing, the coach can see them naked. On that moment he saw two boys on the team circumcised. The first one is staying at Matsapha Police College with both his parents, the other one is staying at Tambuti Estate with his parents also. The first one is 17, he is doing form 5; the other is 20 years old, working at Tambuti estate. The manager said, ‘Oh! I can see now the reason why you are so unfit. You have been busy cutting your penises, forgetting the league is starting and you must go to training. Do you know how much money I have to spend today preparing this game? You were supposed to phone and tell me that you were busy and in pain of these penis-skin cutting thing which will not help you. And you were not going to honor the game because you were unable to even walk , you let the boys pass just next to you. You make me angry, guys.’ He continues, Hey M, why right [now] are you cutting your foreskins? M. answered, this is for protection, babe (term of respect) for viruses. The manager said ‘ nonsense, listen here my boy, I am burying Xhosas almost every weekend, even in midweek because of these knowledge you are carrying on your head. They also think they are safe, but no they are not safe because they are dying of AIDS every day because they are being deceived by those who deceived you.‘ Our manager is staying in Eastern Cape in Mtata, he is a Zionist church leader, he used to come home on Monday but sponsoring our team every time when there is something going on at the team. Our manager continues, ‘guys, please’ in a calm voice, ‘Don’t go and sleep around , you are really going to die. This circumcision thing doesn’t even by a small chance do anything about Aids.’ He continues, ‘Let me tell you guys, this circumcision thing reduces the erection power and that is the only way I can agree that is going to reduce the chances of having you sleeping around with girls, which is where you can get aids.‘ He was on a joking mode, we all laughed, including the opponents, because they were close to us as we were now sharing the shade of the roof of the officials’ house. About a week later, at the next soccer practice, the captain says ‘By the way, are you completely healed, M?’ ‘Yes, you can even ask S., circumcision just takes not more than 6 weeks to heal. I have 23 just circumcised when we closed the league last season. Look now it has been almost three months now.’ S. started and he said ‘Our father (the manager) was just making funny of ourself that day. He knows very well that here in Swaziland we don’t do it like the Xhosas he was talking about. These Xhosa, they go to the mountain where there is no medication at all, and cut their foreskins and not get treatment. If it was so, I would never done it. Something else everybody should know, we must all get circumcised because we really don’t want to see you tomorrow in hospital dying because you are all unstoppable to sex.(S. is our teammate who is also circumcised). The campaign has stimulated a great deal of debate, but no unanimity of opinion. The code for circumcision was applied to all mentions, both positive and negative; conversations where someone asked for information and others provided it were excluded. The community is divided, although there are more negative than positive comments. In any given conversation, a speaker is for it or against, and they always provide a reason (or keep quiet); there are few comments in which a speaker says “I’m for it, but….” or “I’m against it, but….” This is not, of course, to say that in another conversation, with different participants, the speaker might take a different position. Altogether 64 instances of arguments for circumcision, and 90 against were found. A few people were for circumcision in principle, but had some concerns. Figure 6: Frequency of comments for and against male circumcision 100 90 80 70 60 For (64) 50 Against (90) 40 30 For, but with reservations 20 (3) 10 0 Reasons FOR and AGAINST circumcision A study of men’s reasons for refusing circumcision conducted in 2009 in Mbabane, Matsapha and Manzini interviewed recently circumcised men in clinics. The quality of the services offered in government facilities was criticized for long waiting times, rushed services, and unclean syringes and tools (Grund 2010). In everyday talk, however, these issues did not appear to be important with respect to MC. 24 Figure 7: Arguments against male circumcision 16 injury 14 Religion sex bad 12 political Foreskin 10 unSwazi no condom bad 8 pain 6 HIV Unsafe abstinence 4 shame relationships 2 too young too old 0 The most frequent arguments against circumcision were related to concerns about possible injury during the procedure, with its implication for sexual pleasure and performance. Religion is the next most frequent category, followed by concerns about sexual pleasure and sexual performance after healing. Political concerns about the “real” reason for the circumcision drive, as well as concerns about the fate of the removed foreskin (which, although mostly unstated, are probably related to witchcraft) were common, as were comments about the cultural inappropriateness of circumcision (unSwazi). A fair number commented that circumcision would lead to more unprotected (no condom bad) sex with subsequent negative implications for the spread of HIV. The period of required abstinence following the procedure and the possible impact on the man’s relationships concerned a few. Some were ashamed of having undergone circumcision already. This could be because of religious concerns or because of the “unSwazi”ness of the result. Some of the positive comments about circumcision are the mirror image of the negative comments: sex after circumcision is better, there is less risk of HIV and other STIs. 25 Figure 8: Arguments for male circumcision 25 20 HIV Down Sex good 15 HIV Safe STI Down 10 STI Safe Health Good 5 HIV Test Good 0 Key: HIV down: has heard that MC reduces HIV; Sex good: belief that MC enhances sexual prowess; HIV safe: belief that circumcision makes one immune to HIV; STI down: has heard that MC reduces the spread of STIs; Health good: MC generally good for health and personal hygiene; HIV test good: compulsory HIV test is good. Demonstrating sexual prowess is clearly important to men. In one fight, one man accuses another of having many partners, “alternating girls”, but he is not able to satisfy them. “Your wives are hungry for sex. Admit that you are a coward [weakling, impotent]. Polygamy takes a lot of work”. In another argument, a brother accuses his sister’s sugar daddy of having “sugar diabetes”, suggesting that the man is so old he is impotent. Thus, concerns about potency have the potential to dissuade men from seeking circumcision. Conversations about circumcision are more frequent among men than among women: thus, the data are more likely to show whether men think women think circumcision is desirable than to show what women themselves say. Based on the evidence of the journals, there is widespread dissemination of the circumcision campaign information. While it is impossible to tell whether this is due to campaign intervention without similar figures from before the campaign launch, there are many conversations about circumcision, and many references to specific campaigns, events and information gleaned from campaigns (“they say”, I heard”) which suggests that the campaign has stimulated debate and increased knowledge. On the other hand, there is still substantial opposition, in which the implications of circumcision for sexual prowess are a major factor. Sexual concurrency The analysis of the ethnographic journals shows that while there was a great deal of talk about the risks of having multiple partners for an individual’s likelihood of infection, the central claim of the concurrency hypothesis—that compared to serial monogamy, it substantially increases the rate of transmission (Mah & Halperin, 2010) —was not understood. In coding the ethnographic journals, concurrency was defined as any talk that makes it quite clear that an individual was in a sexual relationship with two or more partners at the same time. The coding was difficult, however, since in most of the conversations about multiple partners it was not possible to know whether the partnerships occurred at a different time or were concurrent—itself an indication that the concurrency has not been taken up as an issue. The chosen solution was to restrict the codes to relationships described in the present tense. 26 Thus, any mention of someone with a partner (wife or girlfriend) falling in love with, sleeping with, or mentioning talk by, or about, someone of falling in love with, sleeping with, or mention of having a girlfriend/boyfriend was taken to indicate concurrency. In addition, any indication that the speakers believe that “sharing” partners referred to an overlap of partners, or when it is reasonable to assume that the relationship occurred within a short space of time, is considered concurrency. The exception is for married couples, when one spouse is absent and the other cheats. These were coded for concurrency even though the affair might be short lived and over by the time the spouse returns. The data were coded for the concerns of the speaker over whether it is self or others who would be infected by having multiple and concurrent partners. Figure 9: Negative comments about multiple and concurrent sexual partners 250 comments about infecteing others (87) 200 comments about infecting 150 self (72) self and others (12) 100 Negative comments about 50 concurrency not stated as HIV related 237 0 The data were also coded for whether the concurrent partnership is likely to be short-lived or long-term (or not known). These can only be taken as rough indicators, because several assumptions needed to be made. If someone is described as having a boyfriend or girlfriend while being married or in a stable (boyfriend/girlfriend) relationship, or when someone is described as having an affair, or when anything indicates the affair consists of multiple meetings, the data is coded as “long”. If it is clear that the relationship is a one-off affair, involves paid prostitutes or vulgar talk about having had him/her, it was coded as “short”. If there is nothing to go on, it was coded as “short or long” in the graph. 27 Figure 10: Estimated duration of concurrent partnerships 100 90 80 70 60 50 40 30 20 10 0 Long (87) Short (81) Short or Long (53) This presentation of the way people talk about concurrency is less authoritative than the previous presentation of circumcision. More than half of the respective comments did not mention HIV or AIDS at all, but rather were about other aspects of partnerships such as children, rights of men to have multiple partners, pride in multiple partners, complaints about a cheating partner that indicated that the speaker’s pride was hurt, motives—or imagined motives—for cheating, etc. Furthermore, while there was no ambiguity about what constituted a conversation about circumcision, it was difficult to distinguish between multiple partners in general and concurrency in particular. Even when there was talk of “sharing” partners, it was often not clear whether the sharing was concurrent or sequential. Condoms The approximately equal frequencies of comments favoring condom use and those against it suggest ambivalence in the local communities. 28 Figure 11: Frequency of comments for and against condom use 160 140 120 100 for condom use (133) 80 against condom use (146) 60 40 20 0 Six reasons for condom use could be distinguished, as the following graph shows: Figure 12: Arguments for condom use 70 60 unspecific (51) 50 to prevent HIV (60) 40 to prevent pregnancy (11) 30 sometimes uses condoms (7) 20 STI prevention (2) 10 when cheating only (2) 0 The most frequent reasons given for not using condoms are not new: it reduces sexual pleasure, condoms cannot be used in marriage; it creates suspicion in relationship. A fourth frequent reason, “forgetting when drunk” suggests that men are regularly using condoms when they are sober. 29 Figure 13: Reasons for not using condoms waste of government money(1) intent to infect (1) culturally alien or unaceptable (2) Reasons for not using condoms - Swaziland other person's responsibility (2) already knows or suspects infection (3) prevents pregnancy (3) old, so safe from HIV (3) financial reasons (4) religious reasons (4) shame an embarrassment of acquiring condoms (5) conspiracy (7) creates suspicion in relationships (10) forgetting when drunk (13) not in marriage (15) sexual pleasure (22) 0 10 20 30 40 50 unspecific (49) 30 SECTION 4. CONCLUSIONS Below the findings of the ethnographic study are summarized and conclusions drawn based on the comments collected on various topics and on the reading of the available literature. It is hoped that they can help inform future HIV prevention messaging in Swaziland. 4.1 The value of talk Men and women in local communities in Swaziland seem to enjoy talking about sexual behaviour and sexual partnerships because the talk is often scandalous, and thus fun, and because it allows them to learn what their friends, relatives and neighbors say about issues that are of intense concern, such as AIDS. Because the talk is often personalized—the story of a particular person—the conversations thus permit participants to assess how others would judge their behaviour. Access to the talk in local communities is also valuable to those designing interventions aimed at reducing new HIV infections: interventions tailored to local understandings and meanings are likely to be more effective than interventions that are based on international understandings of how to best combat the epidemic. The amount of critical talk about extramarital partnerships suggests that this behaviour is not publicly accepted. Understanding of the risks of having partners outside wedlock is present in all four research sites (which is consistent with the socio-cultural and epidemic homogeneity of Swaziland, see Whiteside 2005). There is no denial of the danger of death from AIDS, and very little fatalism: of all the comments, fatalistic attitudes were expressed in only 30, most of them from male participants and young participants. Rather, there is criticism of risky sexual behaviour on the grounds that it leads to infection and that it is immoral, both because it is “bad” on a number of dimensions (such as out of wedlock pregnancy) and because infecting others is akin to murder. There is widespread and often harsh criticism of those who “behave badly”. 4.2 Types of partnerships The twelve partnership types which were defined based on the study data ranged from marriage to commercial sex. They were, in order of most frequent to least frequent in people’s comments: a) Single with multiple casual partners b) Married but with a casual partner c) Single with multiple steady partners d) Married but with a steady outside partner who is considered a girlfriend e) Single with both a steady partner and a single casual partner f) Married but with multiple casual partners, married person as promiscuous g) Married h) Polygamy (formal) i) Single but with both a steady partner and multiple casual partners (talk is about having a steady partner and also sleeping around) j) Single with a steady partner k) Married but with multiple steady partners l) Polygamy but with additional multiple casual partners These categories are porous - a casual relationship in which sex is regular can become a live-in relationship, which in turn has the potential to become a marriage. Two types of behaviour are subject to particularly severe criticism: men and women who have multiple partners, and intergenerational relationships. The former are seen as particularly risky and the latter violate local norms. Although there is little social support for intergenerational relationships, even by the youth, these relationships do 31 occur. The participants know they are risky, but young women do not appear to recognize just how risky a relationship with an older wealthy man is. The study found that there is relatively little talk about the risks of HIV transmission in marriage. The application of the MOT model suggested that a high proportion of new infections may occur in people who report having had one sexual partner only during the past 12 months (which includes people in marital union and other stable relationships). Yet this risk is not represented in the frequency distribution of comments about types of partnerships. Rather, people appear to assume that if one spouse is infected the other is also. This is a particularly problematic assumption for men to make. Because women are tested at antenatal clinics, more women are tested than men: the women know their status and the man assumes that his status is the same. 4.3 Motivations for partnerships There is little new in the journals about the basic motivations for partnerships: they are love, money, revenge and sex. Prevention programmes have been particularly sensitive to the importance of transactional sex for women: poor women are seen to be especially vulnerable to HIV infection because they have no alternative other than to supply sex for survival. This is in stark contrast with the talk in the journals about extramarital sex. In the communities, extramarital sex always involves a transaction between a man with money and a woman who wants money. The women are not seen to be practicing survival sex, rather, they are perceived, by both men and women, as mercenary. The comments in the journals make it clear that both men and women understand that men with money are more attractive to mercenary women than poor men: rich men have more partners because they can. 4.4 Cultural scripts and norms Although the 2009 Annual Report of the National HIV and AIDS Response says that “the role of families in instilling moral values around acceptable moral behaviour of their children has been found to be generally weak,” the journals suggest that there is support for norms that restrain what communities consider unacceptable moral behaviour. The norms are more stringent for those who are married, and those who violate them can be punished legally as well as socially. Particularly important for women is that when a woman is married, the foundational assumption is that any child she bears is a child of her husband, who is obligated to support that child and, if she is caring for the child, to support the woman. A married woman can expect to receive social support should her husband abandon her literally or should he abandon her figuratively by spending his money on other women rather than on his wife and his children. Thus, although a high proportion of mothers are not married—the continuation of a pattern set much earlier (Russell 1995, using data collected in 1988-89)—women value marriage. It provides some degree of legal protection for her, she can expect moral and material support not only from her family but from her husband’s, and she can expect moral support from members of her community. Similarly, a married woman can expect more social criticism, and in insulting terms, if she cheats on a husband than if she cheats on a casual partner, and thus may be less likely to cheat than an unmarried woman. The norms regulating women in other types of partnerships, from quasi-marital relationships where the man lives with and supports the woman and their children, to casual partnerships, are weaker. At the far end of the continuum are one-night stands, where there are few, or no, expectations or obligations other than the immediate exchange of sex for money. For married men, marriage is also seen as ideal, in the sense that it is a sign of being adult, a signal to all that he is economically capable of supporting a wife and a child, and that he is a mature person. There is more tolerance for men who delay marriage than for women who do so: in many conversations, 32 a man is justified for “enjoying his youth” until he is middle-aged, when he should no longer be a “youth”. The most stringent norm for men is that a man is expected to support his children. The only socially legitimate justification that he can offer for non-support is that the child is not his. If he is married to the mother of the child, this is a difficult argument to make to a court, to his family and to his community. As the partnership type moves away from marriage, he is more able to wriggle out of claims for support: even so, he may be criticized in public. For participants in conversations, marriage is seen as especially desirable for women, but men are also presented as wanting to be married eventually, even if not just now. The strength of marriage is that a higher standard of behaviour is expected of the married than the unmarried. The more stringent social norms that regulate marriage carry a potential for HIV prevention. A major barrier to marriage is men’s inability to support the costs of marriage. 4.5 Male circumcision The journals show that the offer, desirability and effects of male circumcision have become topics of discussion and debate in every-day life. In the period covered by the journals, the issue of sexual prowess dominates men’s conversations, both as a reason for circumcision and as a reason for objecting to it. Other arguments, such as “circumcision is not Swazi custom”, or “it is not in the Bible”, are less frequent numerically and appear to be called on when the speaker wants to lengthen his list of objections. What is missing from the conversations about circumcision is any indication that participants understand why circumcision is effective. The message that is circulating is simply that experts say it is effective so men should do it. In this sense, it is as mysterious as witchcraft, which the local experts—the elders-- declare is potent—and witchcraft has the advantage that virtually no one argues that it is not a serious issue. It is possible, then, that more information about why circumcision is effective would be useful in persuading men to undergo the procedure (and persuading women to encourage their partners to do so). The health sector has long talked about disease and immunity in terms of the body’s attackers and the body’s soldiers: explaining the role of the Langerhans cells is likely to be no less difficult to translate into everyday speech. 4.6 Sexual concurrency The 2009 Swaziland Prevention Response and Modes of Transmission Analysis points out that the policy environment had not yet responded specifically to MCP as a key driver of the epidemic, and thus that “there is no adequate focus” on this type of partnership (Government of Swaziland and NERCHA, 2009). This is supported by the finding that the central claim of the concurrency hypothesis—that compared to serial monogamy, it substantially increases the risk of acquisition from and transmission to others —has not been incorporated into every day talk. For example, although talk about circumcision referred to the new knowledge that circumcision decreases the risk of HIV, there were no comments recorded of speakers referring to new information about the risks of having more than one sexual partner at the same time. There is certainly much concern about multiple partners, but based on similar concerns in other countries where multiple partners have long been considered risky, it is likely that this preceded the current concurrency campaign. The term “sharing partners” is used, but there is no data suggesting that this is a new expression. 4.7 Misconceptions relating to HIV/AIDS This study identified misconceptions held by the population about serodiscordance, the duration from infection to symptoms, and the correspondence of HIV prevalence and age. 33 Serodiscordance: In the conversations documented in this study, there is an absence of talk about serodiscordance. There is a widespread belief that if one member of a couple is HIV positive the other is necessarily also positive. Because the health facilities require pregnant women to be tested if they want to receive antenatal care, a higher proportion of women in Swaziland are tested than men – in the 2006/07 DHS, 41% of females aged 15-49 years, and only 19% of men aged 15-49 years had ever undergone a HIV test. The combination of a misconception about serodiscordance and the requirement that pregnant women be tested permit men to say that they do not need to be tested, since they will learn their status when their sexual partner is tested. Given men’s fear of testing, promoting couple testing is unlikely to be effective until men have accurate information about the frequency of serodiscordant couples, such that they may make an informed decision. Duration from HIV infection to symptoms of AIDS: In local conversations it is often implicit that speakers consider the duration from infection to symptoms to be much shorter than the epidemiological studies show. The implication of this for prevention is that if a potential partner is known to have previously had a partner who died of AIDS a few years ago but the potential partner still appears healthy, he or she is assumed to be negative. Again, accurate information (on the average duration from infection to symptoms, as well as the variation around that average) would contribute to informed decision-making. Association of HIV with age: There is a lack of knowledge about the association of HIV infection with age for men. It is taken for granted that men with money are more likely to be infected than poor men, and that older men are more likely to have more money than younger men. However, the association of the age of men with their HIV prevalence is not made (although it is evident in the data from the SDHS). A randomized intervention conducted in Kenya found an impact of the simple dissemination of this association to young women in secondary school on the extent of nonmarital, unprotected sex (Dupas 2006; Abdul Latif Jameel Poverty Action Lab 2007). Thus, addressing this issue could be particularly important for young women considering a relationship with a substantially older man. 4.8 In conclusion This study confirmed that hearsay ethnography is a qualitative data collection method which gives insights into sexual and societal norms and cultural scripts held by community members. It provides an opportunity to learn what ordinary people living in the communities in Swaziland say to each other— rather than to an interviewer or focus group moderator-- about relationships between the sexes and norms attached to different sexual behaviours. It captures both spontaneous responses to the epidemic, and responses to prevention messages and programmes. It is therefore an additional data source which can complement data from interviews of informants and from focus group discussions, in order to inform communication aspects of HIV prevention programs. 34 BIBLIOGRAPHY Aral, Sevgi O. 2010. Partner Concurrency and the STD/HIV Epidemic. Infectious Diseases DOI 10.1007/s11908-010-0087-2. Boily, Marie-Claude. 2010. Polygyny, concurrency, its impact and lack of impact on HIV. HIV Therapy 4, no. 2: 139-44. Central Statistical Office, and Macro International . 2008. Swaziland Demographic and Health Survey 2006-07, Calverton, Maryland. Colvin, M. , M. Gorgens-Albino, and S. Kasedde. N.D. Analysis of HIV Prevention Response and Modes of HIV Transmission. The UNAIDS-GAMET Supported Synthesis Process. De Walque, Damien. 2007. Sero-Discordant Couples in Five African Countries: Implications for Prevention Strategies. Population and Development Review 33(3):501-23. Desmond, Nicola, Caroline F. Allen, Simon Clift, Butolwa Justine, Joseph Mzugu, Mary L. Plummer, Deborah Watson-Jones, and David A. Ross. 2005. A typology of groups at risk of HIV/STI in a gold mining town in north-western Tanzania . Social Science and Medicine 60: 1739-49. Dunkle, Kristen L., Rachel K. Jewkes, Heather C. Brown, Glenda E. Gray, James A. McIntyre, and Siobán D. Harlow. 2004. Transactional sex among women in Soweto, South Africa: prevalence, risk factors and association with HIV infection. Social Science & Medicine: 1581-92. Fox, Ashley M. 2010. Survival Sex or Consumption Sex? Gender, Wealth and HIV Infection in 16 sub-Saharan Countries. Ms. Garnett, Geoffrey, Simon Gregson and Karen Stanecki. 2006. Criteria for Detecting and Understanding Changes in the Risk of HIV Infection at a National Level in Generalized Epidemics. Sexually Transmitted Infections 82: 48-51. Gorgens, Marelize. 2010. Evaluation of HIV Prevention Programmes: How Do We Know We’ve Been Successful? Powerpoint presentation. Government of the Kingdom of Swaziland, and National Emergency Council on HIV and AIDS (NERCHA). 2009. Swaziland: HIV Prevention Response and Modes of Transmission Analysis, NERCHA, Mbabane, Swaziland. Grund, Jonathan. "A Cut Above the Rest." Available at http://www.iavireport.org/archives/2009/Pages/IAVI-Report-13%286%29- cut-above-the-rest.aspx?utm_source=iavireport&utm_medium=email&utm_campaign=IRNovDec2009. Grund, Jonathan. 2010. Perceptions of Risk and Sexual Behaviour Change following Adult Male Circumcision in Urban Swaziland. MPH Dissertation, Hubert Department of Global Health, Emory University. Halperin, Daniel, and Helen Epstein. 2004. Concurrent sexual partnerships help to explain Africa's high HIV prevalence: implications for prevention. The Lancet 364: 4-5. Halperin, Daniel T., and Helen Epstein. 2007. Why is HIV Prevalence so Severe in Southern Africa? The role of multiple concurrent partnerships and lack of male circumcision: Implications for AIDS prevention. The Southern African Journal of HIV Medicine, no. March: 19-25. Hunter, Mark. 2010. Love in the Time of AIDS. Bloomington, IN: Indiana University Press. Hunter, Mark. 2007. The changing political economy of sex in South Africa: The significance of unemployment and inequalities in the scale of the AIDS epidemic. Social Science and Medicine 64: 689-700. Hunter, Mark. 2005. Cultural Politics and Masculinities: Multiple Partners in Historical Perspective in Kwa-Zulu Natal. Culture, Health and Sexuality 7 (3): 1-17. Kingdom of Swaziland, and NERCHA. 2009. National HIV and AIDS Response. NERCHA, Mbabane, Swaziland. Kingdom of Swaziland and NERCHA. 2009. The National Multisectorial Strategic Framework for HIV and AIDS 2009- 2014,.NERCHA , Mbabane, Swaziland. . Kratz, Corinne A. 2010. In and Out of Focus. American Ethnologist 37, no. 4: 805-26. 35 Kretzchmar, Mirjam, Richard G. White, and Michel Caraël. 2009. Concurrency is more complicated than it seems. AIDS (online preprint). Kurimoto, Nami. 2011. HIV Testing among Married Couples in Swaziland: Social and Economic Determinants and Associations with Condom Use and Fidelity. PhD Dissertation, Department of International Health and Development, School of Public Health and Tropical Medicine, Tulane University. Leclerc-Madlala, Suzanne, Leickness C. Simbayi, and Allanise Clloete. 2009. The Socio-Cultural Aspects of AIDS in South Africa . HIV/AIDS in South Africa 25 Years On: Psychosocial Perspectives . Eds Poul Rohleder, Leslie Swartz, Seth C. Kalichman, and Leickness C. Simbayi, 13-26. Springer. Lurie, Mark N., and Samantha Rosenthal. 2009. The Concurrency Hypothesis in Sub-Saharan Africa: Convincing Empirical Evidence is Still Lacking. Response to Mah and Halperin, Epstein, and Morris. AIDS Behaviour (online preprint). Mah, Timothy L., and Daniel T. Halperin. 2010. Concurrent Sexual Partnerships and the HIV Epidemics in Africa: Evidence to Move Forward. AIDS Behaviour, 14(1):11-16. Morris, M., and M. Kretzshmar. 1995. Concurrent partnerships and transmission dynamics in networks. Social Networks 17: 299- 318. Padian, Nancy S., Charles B. Holmes, Sandra I. McCoy, Rob Lyeta, Paul D. Bouey, and Eric P. Goosby. 2011.. Implementation Science for the US President’s Emergency Plan for AIDS Relief (PEPFAR). Journal of Acquired Immune Deficiency Syndrome 56, no. 3: 199-2003. Reniers, Georges, and Susan C. Watkins. 2010. Polygyny and the spread of HIV in sub-Saharan Africa: a case of benign concurrency. AIDS 24: 299-307. Ruark, Allison Herling. 2008. HIV/AIDS Risk and Marriage in Africa. Paper presented at a Global Health Council Annual Conference on Faithfulness and HIV Risk Within Marriage: Programme and Research Perspectives, 28 May, 2008. Russell, Margo. 1995. Women, Children and Marriage in Swaziland . In Woman and African Society . Eds Man Sing Das, and Vijay Kumar Gupta, 1-15. New Delhi: M D Publications Pvt Ltd. SADC. 2006. Report of Expert Think Tank Meeting on HIV Prevention in High HIV Prevalence Countries in Southern Africa, 10-12 May, Maseru, Lesotho. Sawers, Larry, and Eileen Stillwaggon. 2010 . Journal of the International AIDS Society 13, no. 34. Shelton, James D. 2009. Why Multiple Sex Partners. Smith, Daniel Jordan. 2010. Promiscuous Girls, Good Wives, and Cheating Husbands: Gender Inequality, Transitions to Marriage, and Infidelity in Southeastern Nigeria. Anthropological Quarterly 83, no. 1: 123-52. The Kingdom of Swaziland. 2010. Swaziland Estimates and Projections. Mbabane: NERCHA and UNAIDS. Swidler, Ann, and Susan Watkins. 2010. Practices of Deliberation in Rural Malawi. Paper presented at the conference on Deliberation for Development: New Directions. Washington D.C., November 12 & 13. Swidler, Ann, and Susan Cotts Watkins. 2007. Ties of Dependence: AIDS and Transactional Sex in Rural Malawi. Studies in Family Planning 38, no. 3: 147-62. The Government of the Kingdom of Swaziland, and The National Emergency Council on HIV and AIDS (NERCHA). 2009. Swaziland HIV Prevention Response and Modes of Transmission Analysis. NERCHA: Mbabane, Swaziland. Van de Walle, Etienne. 1993. Recent Trends in Marriage Ages. In Demographic Change in Sub-Saharan Africa. Washington, D.C.: The National Academics Press. Watkins, Susan. 2004. Navigating AIDS in Rural Malawi. Population and Development Review 30 (4): 673-705. Watkins, Susan C., Ann Swidler, and Crystal Biruk. 2011. Hearsay ethnography: A method for learning about responses to health interventions. In The Handbook of the Sociology of Health, Illness and Healing. Eds B. Pescosolido, J. McLeod, J. Martin, and A. Rogers, 431-45. New York: Springer. 36 Whiteside, Alan with Alison Hickey, Nkosinathi Ngcobo and Jane Tomlinson. 2003. What is driving the HIV/AIDS epidemic in Swaziland and what more can we do about it? National Emergency Response Committee on AIDS and United Nations Programme on HIV/AIDS, Mbabane, Swaziland. Whiteside, Alan. 2005. The Economic, Social and Political Drivers of the AIDS Epidemic in Swaziland: A Case Study. In The African State and the AIDS Crisis, ed. Amy S. Patterson, 97-124. Aldershot, England: Ashgate Publishing Limited. 37 Annex 1: Data and Methods The credibility of any research depends on evaluating the methods by which the data were collected and analyzed. Below we present these in some detail (for more details, see Watkins and Swidler 2009). Ethnography is a staple of anthropology. The advantage of ethnography over other forms of qualitative data collection is that it provides information on a community, rather than on individuals, and thus provides information on cultural scripts and norms that are external to individuals and expressed through informal and social interactions. In the classical version of ethnography, a single person, usually an outsider spends a year or more in a community, learning the language, observing and listening, and then writing field journals (diaries) detailing what he/she observed and heard. This approach is quite time-intensive and limited to one site: thus, it is not suitable for the current assignment. Selection of Sites: Preparations for the study began in September, 2009, when four study sites were selected by NERCHA. Among the selection criteria were: 1) that the site should not be fully rural, since due to the spread-out nature of the homesteads, there was less likely that there would be frequent public conversations available to the ethnographers; 2) that the site should not be in one of the two major cities, Mbabane, the capital, and Manzini, a large commercial center, or the corridor that links the two cities; 3) that the distribution of sites capture some of the diversity in the country: thus selected sites included Ludzeludze, Bhuleni, Siphofaneni and Mathandele. The selected sites had a common characteristic of being areas of high development or showing a potential of high development. Buhleni and Siphofaneni are “rural growth points”, i.e. rapidly growing and urbanized areas. Ludzeludze is not far from Manzini, and people go back and forth frequently, suggesting considerable urban influence, and Mathandele is under the jurisdiction of the Nhlangano Town Council, and is considered urban. Selection of ethnographers: In consultation with others, it was advised that the initial recruitment of potential ethnographers in these sites be made by the clerk of the Kagogo Centre in each site. In three of the sites, the study team (Watkins plus Assistant Researchers Lucas Jele and Xoli Sereo from NERCHA) then met with the clerk of the KaGogo Centre there, described the study and the method, and asked him/her to recruit 10 potential ethnographers who were high school graduates and could write in English. Since Mathendele is under the jurisdiction of Nhlangano Town council, the site does not have Kagogo clerk, and the selection was made by the HIV prevention coordination office of the Nhlangano Town council. It was not expected that the selection of the ethnographers would be based on ability alone. Jobs for high-school graduates in the study sites are very scarce and thus a valuable commodity. The clerks surely took into account that this was an opportunity to distribute patronage. This assumption turned out to be correct; in one site, most of the ethnographers had the same surname as the clerk and others might have been relatives as well. In another site several recruits were manifestly unsuitable— one told us he spent much of his time at the shebeens. He seemed to have some responsibilities at the centre but was probably paid a pittance, so the clerk may have felt that this would be a way to reward him. In another site, several of the recruits were members of the same church as the clerk. The clerks were asked to recruit more ethnographers for training than would be needed, since earlier experiences with Hearsay Ethnography in Malawi and Botswana showed that some of those recruited would not be suited to the task and some would drop out, probably because they did not enjoy the work, it was not sufficiently well compensated, or they left the area. In Swaziland, 33 potential ethnographers were recruited and trained. Characteristics of the Ethnographers: They are quite a diverse group in terms of their demographic characteristics. This diversity does not mean that the ethnographers were representative of the diversity of their settings. However, since an individual’s social networks are typically homophilous, this suggests that the diversity of the ethnographers means that they will go to a variety of places, and thus that the conversations reported in the journals, came from many different segments of the population. Although the coding of the journals (described below) did not include coding for the characteristics of the speakers (e.g. age, gender), this could be done. 38 Of the 33 ethnographers, most (19) had completed their 0 levels (Form 5), three had only completed Form 3, three had only completed Form 4, and the rest had some sort of diploma, e.g. in accounting. Their last year of schooling, which is an indicator of their age, their varied substantially: three completed schooling in the 1970s (the earliest in 1978), one in the 1980s, and the others more recently. Only 10 out of the 33 were married, consistent with the late age of marriage and the high proportion of those who do not marry, in Swaziland. Many of those who said they were single had at least one child. Most (22) were unemployed, three were KaGogo Centre clerks (i.e. they selected themselves), one was a pastor, and the others who did not say they were unemployed had jobs that likely paid very little, such as a cook and an assistant cook at a feeding center for orphans. On religion, all said they were Christian except one, who said he was “Swazi traditional religion.” All said their first language was SiSwati, their second was English (although three, all older women, were sufficiently weak in English that we permitted them to write their journals in SiSwati). (For the full demographic details, see Appendix 1). Ethical Approval of the study: Ethics approval was given by the Scientific and Ethics Committee of the Ministry of Health and Social Welfare and written comments received. A response was submitted in writing, and the lead consultant met with the Committee, where additional issues were discussed. Two categories of questions dominated both the Committee’s written and verbal comments. One set required the researchers to explain more clearly the differences between Hearsay Ethnography and surveys in terms of representativeness and data quality; the second set concerned issues of confidentiality and consent. From these questions, it was clear that it had not been made sufficiently clear by the researchers that they were only interested in public speech, not private, and that this speech could not be interpret in terms of actual behaviour. Once this was explained, the Committee granted approval for the study. Training The study team spent three days in each site training the recruits. The training emphasized several key aspects of the methodology: the ethnographer was not to interview, but simply to listen to public conversations about AIDS, defined as conversations that other people could overhear, e.g. on a bus, in a shebeen, at a football game, etc. The ethnographer was not to alter his or her daily routines or to go to special places. It is reasonable to assume, however, that when an ethnographer had not “filled” his journal shortly before the Assistant Researcher came to collect the journal and pay the ethnographer, that they did go to places where they thought they would find people talking in public. The definition of “a conversation about AIDS” was left up to the journalists. Almost invariably, the conversations recorded in the journals had to do with the heterosexual transmission of HIV or deaths from AIDS, although there were also some on antiretrovirals and on orphans. The training was a combination of teaching and practice, with the mornings devoted to teaching and the afternoons and evenings to practice. On the first day, we introduced the project for about an hour, including the basics of journal writing: to capture in detail the direct speech of those who were talking, and not to summarize. We then gave them a practice journal and told them to find a place to sit alone and for half an hour to write down, in as much as detail as possible, everything that had been said. Rarely could they write more than a page. When they returned, we read them aloud, so each could see what he/she missed that another recruit had captured, and that he/she was no worse than the others. As we read, we pointed out where they summarized, what detail was missing. After this exercise, we went through the do’s and dont’s of ethnography in detail. (See Appendix 2). Then the ethnographers were dismissed, and asked to return the next day with a journal entry. The teaching emphasized that the aim of the research was to produce accurate and detailed accounts of the ways people talk to each other in public and in everyday contexts. This would permit NERCHA and others to understand the people share information, the debates they have about how to make sense of the forces acting on their lives, and the stories they tell each other about how life should be lived. An effective way of getting this across was to ask them to think of themselves as video cameras, simply recording in their mind what they heard and saw, and then translating that into a journal account. This 39 made it clear that the research was not about either the ethnographer’s thoughts and behaviour nor the behaviour of those who were speaking in public. Emphasis was also placed on capturing, as much as possible, the exact words of the speakers rather than summarizing, i.e. “He said ‘The more girlfriends the better, that way I can have my choice’” rather than “He talked about having many partners.” During the conversation, no notes were to be taken. Afterwards, the ethnographer could, in a private place, jot down a few notes which would serve as reminders when he/she wrote up the encounter in the evening. The journal should be updated as soon as possible, while the interaction was still fresh in the ethnographer’s memory. Inevitably, however, some conversations were recorded by the ethnographers sometimes many hours after they had heard the original conversation, and some of the natural ebb and flow of the conversation would have been sacrificed in order to record the relevant content. The journalists were expected to write in English. Two exceptions were made. First, there were three older women who asked to write in SiSwati. They were permitted to do so because their activities were likely to bring them in contact with conversations among older people, thus adding diversity to the ethnographic study. The second exception was for instances in which the ethnographer was not certain how to translate a phrase, such as a proverb. In such cases, the ethnographer was to write the phrase in English and then try to translate it. The journals were to be kept in a private place. The ethnographers were encouraged to use pseudonyms if they happened to know the person/persons speaking. If this was too difficult, the ethnographers should be absolutely sure that no one else would have access to the journals, since it would be unethical to do otherwise. In training, all said that they had a private place where they could keep the journals. At the end of the training each journalist was given pens and a hard-backed journal with 80 pages, which they numbered. They were to be paid approximately US$20 (E850.00) for a full journal that was acceptable to the Assistant Researcher who collected it, although they could also submit half a journal and be paid half rate (E425.00) for it. The journals were collected monthly and the journalists were paid at that time. An important aspect of the training is that it continues beyond the formal training. The most common difficulties that the ethnographers have had are 1) capturing direct speech rather than summarizing; 2) putting themselves in the picture as enlightened about AIDS—for example, expressing their approval of those who speak approvingly of condoms, or their disapproval of those who speak enthusiastically about the benefits of having many partners. A third problem is making things up: this happened rarely, perhaps because it is more difficult to make up a conversation than to recall one. To ensure that these problems do not persist, it is quite important that the person who collects the journals monthly should review at least two excerpts, and where it is evident that the ethnographer had not understood the instructions, to provide suggestions for improvement. In addition, the journals should be sent quickly to the Research Director (in this case Watkins) who could then provide suggestions for improvement that could be given to the ethnographer by phone or at the latest at the time of the next journal collection. If the ethnographers do not continue to receive training after the study has begun, as was the case with this study, it is likely that some of the journals will be inadequate. Several ethnographers did not do well enough in the training to continue; several did not submit any journals at all or submitted only a few. Two journalists plagiarized and were fired; their journals are not used in the analysis. Data collection: Some ethnographers produced one journal for each of the six months of the study, some produced only one. The mode was 4 journals/ethnographer. A total of 104 journals were submitted and paid for. Of these, 12 were discarded. This includes all the journals of the two journalists who were fired for plagiarism, as well as some journals that Crouse and Watkins thought were not convincing. There were also parts of otherwise-acceptable journals that appeared to us to have been fabricated: the mode of speech reported in the journal was different from the ethnographer’s other writing, and from the mode of speech in the journals of the other ethnographers. These parts, amounting to approximately 3 journals, were also ignored in the analysis. 40 There were a total of 992 conversations about AIDS. Some of these were quite long (15 pages), others were brief. The ethnographers had been asked to describe the context in which the conversations occurred, including what the participants were wearing (to provide a sense of their economic status). Often many pages were written on context and clothing; these are not included in the count of the pages of each conversation. Coding Coding ethnographic texts is quite different from coding survey data or a structured in-depth interview or focus group. A conversation may start with someone on a bus waiting to depart commenting on the age discrepancy of a couple walking next to the bus, this may lead to talking about mercenary young women or to men who are behaving inappropriately by going for someone the age of their daughter; someone on the bus may chime in, sending the conversation in a related but different direction. In addition, some of the issues that are relevant for this study are implicit. For example, a conversation about women who wear miniskirts to seduce men may not mention AIDS, but everyone will have heard other conversations, or read scandals in the newspapers, that linked such women to the spread of AIDS, whether or not Aids is explicitly mentioned and whether or not “unprotected sex” is mentioned. The coding scheme was developed by Watkins, and then revised in consultation with Jele and Crouse. All the journals were coded by Heinrich Crouse, who was selected by the World Bank. Crouse has an academic background in social anthropology and has experience in qualitative research and analysis. He consulted by email with Watkins initially, and then worked independently, with occasional email discussions with Watkins when either raised a question about the quality of a journal or about the veracity of a particular excerpt. The coding required intensive reading. Because of the ambiguities of ordinary conversation, it was often necessary to read an excerpt several times, sometimes adding or deleting codes with each successive reading. This was particularly the case since English is not a first language for any of the ethnographers: when ethnographers were uncertain, they often included the original dialogue in siSwati. In addition, , the handwriting of some was very poor. It thus often took several readings to be able to identify relevant content. The initial coding of a journal took four hours. For both Crouse and the Swazi coders, the instructions were to apply the 18 codes to conversations that the various ethnographers recorded in order to make some kind of quantitative analysis of qualitative data. These codes identified fights and arguments, relationship types, motivations for engaging in sexual relationships (men and women), responsibilities of men and women in sexual relationships, and prevention themes (which included mention of condoms, circumcision, concurrent and multiple partners, intergenerational sex, clinics, alcohol and stigma). Talk of witchcraft was coded as well as any insults. (For a detailed list of the codes, see Appendix 3). Intercoder reliability: In addition, some of the journals were coded by five coders selected by NERCHA (Jele), in order to permit assessing intercoder reliability. The criteria for selecting the Swazi coders included recognizing the qualities that are needed to undertake the coding job. The following qualities were requirements from the coders; 1) Health teaching/programming background, and/or 2) Research teaching/conducting background. Jele reviewed a database of researchers who attended a research meeting after the HIV prevention indaba in November 2010 at the Convention Centre. From the research participants attendance list, he acquired the contacts for Dr S. Sithole and Hlobsile Nkambule who are lecturers with the university of Swaziland in the faculties of Health and Social Science respectively. He then sent SMS messages to these two participants requesting to meet them to talk about possibilities of their participating in the ethnography research study; they replied and agreed to be involved in the study, following which formal arrangements with the University were made. Dr Sithole recommended to colleagues, Siphiwe Thwala and Dr Mkhonta, both of whom are interested in research and have experience of conducting ethnography research (for Siphiwe) and qualitative research (Dr Mkhonta). There was also a need to balance the academicians coding point of view with a programming point of view and Thabile Maripe was contacted to fill in that role. Coding qualitative data is difficult, and perfect agreement is not expected. In this case, the intercoder reliability was quite satisfactory. Some previously missed code-able content was indeed identified, but 41 this amounted to less than 5% of the original codes. None of the local coders that were used to check coding reliability identified more code-able content than the main coder used, and some significantly less, but having the main coder work with a local coder lead to a significant increase in code-able content. This exercise clearly showed that best results can be obtained by having 2 coders, at least one of which is a local language speaker, code a journal at the same time. This encourages debate and clearer understanding of the content of a conversation, minimized fatigue errors and of course, makes local language content accessible. Coding process: The journals were scanned and emailed to the coder. Using features of Adobe Pro, the coder highlighted relevant conversational extracts, and wrote codes on the body of the text, using a transparency setting that allows the text underneath to still be visible. These entries were then copied to an Excel document for each journal to make counting and future manipulation easier. Both the journal (now with codes and notes) and the Excel document was then sent to Watkins for checking and to Cristina Romero for filing and storage at the World Bank. The journals were identified by country, town, surname and name of ethnographer, date range of the journal (e.g. Recorded from 101205-101215), journal number (e.g. 3rd or 4th journal received from ethnographer), and coder’s initials. During the coding of the first journals, the coder referred continuously to the detailed coding instruction received to satisfy him that the codes were being applied correctly. Because of the immense range of relevant content, it was not always easy to immediately assign content to a particular code. The clinics code (4e) for instance, was used to capture content not related only to talk about clinics, but also about ARV protocol, traditional healers and PLWA. To facilitate future analysis and deeper coding, the coder used not only the codes, but brief, one sentence descriptions of the scenario which typically identify the gender of the speakers and main themes of the excerpt. Assessing data quality: While there are several techniques for assessing the quality of survey data (e.g. the same question asked in different parts of the survey, the same question asked on multiple surveys, the use of biomarkers to detect whether those who claimed to be virgins had a sexually transmitted infection), there are few for assessing the quality of qualitative data. One approach is to compare responses across different modes of data collection. Phase 1 of this study was to include in-depth interviews, but it proved to be impossible to conduct these within the time allotted. An assessment of the quality of the ethnographic data can be based on comparisons across and within journals. Reading multiple journals, it becomes evident what a normal conversation in this context is like. For example, there are excerpts that are convincingly real, such as the interaction between the bank teller and his girlfriend, or the interactions between the husband, wife, girlfriend and onlookers in the excerpt in the court. There are other conversations reported in a journal that are not credible. For example, the excerpt may describe one speaker asking another “what is circumcision?” a second answers, the first then asks “why is this important?” etc. Such stilted interactions as so rare in the journals that they are not credible as informal conversations. These questionable excerpts were often at the end of a journal, suggesting that the ethnographer had not completed his journal when the due date for collection and payment approached, and simply tried to fill up pages. The excerpts that lacked credibility were not used in the analysis that follows. More serious issues became evident during the coding. One was the importance of excluding spurious conversations from the analysis. Such conversations were marked by, inter alia, highly formal modes of dialogue that are markedly different from the rest of the recorded conversations; atypically fact-based conversations which often take an question-and-answer format with information that seems often to derive from HIV/AIDS awareness literature; conversations in which the ethnographer could not possibly have been present (for example describing the words and actions of people in two different locations), and any scenarios that raised serious doubt in the mind of the coder. Moreover, some entries were also repeated in 2 or more journals. Once this problem was identified, the coder would read multiple journals from the same ethnographer in succession, in order to pick up this kind of repetition more easily. These passages, like the conversations that appeared spurious, were isolated and not coded. 42 Counting: It is not surprising to find that in ordinary conversations, people will repeat themselves. This presents a potential bias. To avoid this, Crouse developed the following protocol was followed: If the person presents the same argument against, say, circumcision several times in the same conversation, it was coded for only once. If they presented different arguments against circumcision, it was coded once for each of those reasons. If they present the same argument in different contexts, each conversation with different participants was coded separately. The rationale is related to the interest in the circulation of cultural scripts. The more a script is expressed, all else being equal, the more effect we can assume it has, and the heavier it should be presented in the coding. Having said this, in rare situations an ethnographer might describe a regular cast of characters and some of these characters might feel strongly about a particular topic and issue opinions and arguments about it on a regular basis. In these isolated cases, heavier coding was used for such a person for the same reason stated above – it was assumed that, all else being equal, the effect of this person’s opinions had more of an effect on the knowledge and opinions of others than someone who might say the same thing only once. Care was however taken not to give too much weight to one person’s input. After coding, entries for each code were collated to produce a document that contains all of the instances of that code, and identifies the journal and page number where the original entry can be found. For some of the codes, the original extracts were also copied and pasted to the same Excel document. 43 Annex 2: Hearsay Ethnography Training Notes for Ethnography Recruits & Suggestions for Trainers 1. Background Hearsay ethnography is a method for studying conversations and social interactions in their “natural” social settings—for example, while waiting for a bus or in a store or at a football game or a shabeen. Hearsay ethnography aims to produce accurate and detailed accounts of the ways people talk to each other in public and in everyday contexts. From this, we can study the ways people share information, the debates they have about how to make sense of the forces acting on their lives, and the stories they tell each other about how life should be lived. We are not interested in what the ethnographers themselves say or what they do in their private lives. Think of them more as video cameras, simply recording the places where people talk and what they say. We are also not interested in any particular people’s private lives—if what the ethnographers hear is gossip about someone’s infidelity, for example, we don’t want to know about the infidelity itself but rather what the people in the conversation say about it. 2. Qualifications Who would be a good ethnographer? This would be i) Someone who can just listen. A good ethnographer does not voice his/her own opinions, but rather is interested in what others are saying. We don’t care, for example, if the ethnographer thinks multiple concurrent partners are a driver of the epidemic, but rather how people talk about people who do have multiple partners, the detailed stories they tell about them. ii) Someone who gets around rather than someone who spends all day at home. “Gets around” means things like going to the store, getting water from a standpipe, taking a bus trip to a market centre, etc. iii) Someone who can write well enough in English that we can read their journals in English. The grammar does not have to be good, just clear enough to be understood. People with less than Form 3 are probably not going to be able to do this. Many social scientists studying AIDS are convinced that the ways people talk together in everyday contexts are the most important influences on the ways they make decisions about actions, such as unsafe sex or having multiple partners, that might expose them to HIV. Most social science methods, however, set up an artificial context – typically an interview or focus group discussion - in order to ask people to speak about things such as their experiences and knowledge of the disease, their sexual behaviour, or their attitudes to illness. While these methods can produce valuable information, what they can not do is reveal how ordinary people talk to each other about these issues on their own and in the places they call their own – their homes, streets, workplaces, shebeens, churches, buses and everywhere else that people gather to chat in their everyday lives. HEARSAY ETHNOGRAPHY TRAINING SUGGESTIONS FOR TRAINERS Day 1: After welcoming the potential ethnographers, go over the Hearsay Training Notes. Then have a snack break. After that, have the ethnography recruits go outside and write up what had been said in the meeting before the break, such as the setting and who said what to whom. Emphasize that they are to itemize, not summarize—we want direct speech, in quotation marks, the actual words the person said. 44 When they return, read a few of them aloud, so they can see what others wrote that they missed (or what others missed that they wrote). Then, after lunch, give them practice journals (the school type) and an assignment: do whatever you would usually do in the afternoon, and write whatever conversation you heard in your practice journal. If they do not hear a conversation about AIDS, they should write about any public conversation they hear, so that we can see if understood the basic instructions (itemize, direct speech, describe the setting and the people, etc). Day 2: After a bit of informal talk about their experiences, have each one read his/her journal excerpt aloud, and ask the others for comments—did they describe the setting well enough so that a reader can visualize it? Did they provide direct speech or did they summarize? Then the ethnographers are dismissed with a longer assignment: to write at least 10 pages over the next three days. [a 3 day break while the trainer starts at another site] Day 3: Again, have them read their journal entry aloud (or, if it is long, only a part of it). Then take a snack break, during which any of the recruits who have not understood the instructions should be taken aside and released. The others are to be given hardbound journals, a copy of the training notes, and explained that they would be paid for a full journal (in this case, 80 pages, the size of the bound notebooks). Before they left, they should number the pages (in Botswana, some handed in journals with pages removed). Emphasize again the importance of writing something every day—that if they waited until a few days had passed after a conversation, they would forget, and wouldn’t get the direct speech. Before dismissing them, get their contact information and give them the date when the first journal will be collected. If there is a contract to be signed, it should be signed at this point. In Botswana, we collected the journals and paid them once a month. In Swaziland, we collected the lst 40 pages after two weeks, and then the 2nd 40 pages: this was to motivate them—they would be getting the money sooner. And for those that didn’t write regularly in their journal, it was easier to catch up for a two week period than a one month period. In Botswana, the Research Assistant who collected the journals read at least two of the excerpts while he was there, to make sure that the person was doing o.k., and to catch people who were just making things up. In Swaziland this did not happen, and it was a problem—by the time I got the journals and read them, those who had done poorly had done another journal. It’s very important for the PI to read the journals promptly and to provide feedback—praise and gentle criticism. This will let them know that you care about their work—and if they don’t have that sense, they do not do as well. Having the journals typed makes subsequent coding much easier, but it is possible to scan the journals instead and to code directly onto the scan using Acrobat Professional. 45 Annex 3: Coding protocols (Heinrich Crouse) The aim of the coding is to be able to provide some quantitative information that will complement the qualitative analysis. Below I begin with a general summary of the sort of overall questions we are asking, and then suggest some specific topics with associated codes. The texts to be coded are reports by ethnographers of conversations they overheard or participated in that have something to do with AIDS. The ethnographers kept journals in which they wrote what they heard each day (although sometimes they skip days if they didn’t hear anything). Thus, the organization of the journals is by excerpts, usually between 2-5 pages, each with a date; it is easy to see where an excerpt begins and where it ends. The ethnographers were asked to begin each excerpt with a description of the setting, who was there, information about the age of the participants and how they were dressed (in order to give a sense of whether this was a relatively wealthy or a relatively poor person), whether the ethnographer knew the person and, if so, whether the ethnographer knew the person’s level of education. and, if the ethnographer knew the person. Typically an excerpt covers one topic, but sometimes there are several topics that are relevant for us. This information does not need to be coded. Coding Procedures: The texts to be coded are in PDF form, so we can’t use a qualitative analysis programme such as Nvivo. We can, however, make use of the features of Acrobat. These permit typing on a PDF, highlighting sections of the text, and adding comments. In principle, what you would do is turn on the “highlight” function to mark a section of the text (not just the line where a particular word like “circumcision” occurs, but the sentences leading up to that and following it) and then turn on the “typewriter” function to type in the code for that section; if you have questions or are uncertain, just insert a comment. It will take a bit of experimenting, but I think this will work well. I have only been able to highlight in one color, yellow, but my guess is that you might be able to find a way to get other colors: then, sections of the text that are relevant for more than one code can be done in different colors, which will help in the analysis. It will take some time to become familiar with the way people talk about AIDS in Botswana and Swaziland, but after a while the coding will be much easier. Specific Codes 1. Fights (this code word includes quarrels). We usually only know about these when they are public: those fighting—often a wife and a girlfriend, but sometimes a husband and a wife or two men— shout or cry “help me help me” and other people gather around to watch, listen, and call the police. Sometimes, however, people will talk about a fight that they saw earlier or heard about. For fights/quarrels, highlight from the beginning of the fight/quarrel to the end. a. Violent b. Divorce, “fired”, kicked out. 2. Partnerships. This is a favorite topic, so there will be lots. There are two things we are most interested in re partnerships: i) the type of partnership, from very casual to legal marriage and ii) the motivation for the partnership and iii) responsibilities. a. Types of partnerships: The distinction between casual and marital relationships is very important, so there’s a separate code for each. i. Casual—this includes talk about people who have many partners, people who change partners frequently, dumping partners. There are two main types of this talk: one is critical (why people shouldn’t have many partners, why they shouldn’t change partners frequently, etc) and comments about how it is good to have many partners (this is usually from men). 46 ii. Marriage (color and bold)—it is sometimes hard to know whether a partnership is a legal marriage (with lobala or a western wedding) or whether it is a steady relationship (the couple is living together and have a kid), so both should be included under this code. It includes talk about wanting to get married (or not wanting to get married), mention of whether the marriage is legal or not, concerns about a cheating spouse, belief that a spouse is faithful or innocent, comments that someone is getting old so he or she should get married, comments about the relatives interfering or supporting a marriage, and responsibilities of marriage. b. Motivations for partnerships: the main ones have to do with love or money , and these are usually talked about as quite different for men and women, so I’ve given them different codes i. Motivations-women. Usually, women in a relationship or contemplating one will say it’s all about love. But when others are commenting about women, will say it’s about money, that women are mercenary these days. Sometimes a woman, however, will refuse a proposal, saying “I have my own money.” Use the same code for both motivations for women, as well as a combination of love and money. ii. Motivations—men “Love” enters less into conversations about men’s motivations, except when the man is making a direct proposal, saying to a woman he has just met “I love you, I want to marry you” . But also use this code for men wanting to settle down or being told they should settle down, or men claiming that they don’t want to settle down they want to enjoy their youth. Also include comments on men with or without money-- how men with money spend it on girlfriends (sometimes this is positive, it’s what men should do, but other times it will be that men who spend money on girlfriends are depriving their wife and children); “men without money” includes comments on not having enough money to have a girlfriend or support a wife or child, comments about how there is no employment these days. c. Responsibilities of each type of partnership i. Women 1. Marriage 2. Casual ii. Men 1. Marriage 2. Casual 4. Prevention themes: All of these get the same color highlighting a) Condoms b) Circumcision c) Concurrent partnerships (different from multiple partnerships—for this code, there has to be some mention of the terms of the prevention campaigns such as “duplicating partners” or “a chain” or “you can’t know who he/she slept with before”, sharing girlfriends) d) Intergenerational 47 e) Clinics—include here testing, e.g. “know your status” or “I am not ready to know my status”; anything on ARVs (“she swallows tablets”, “when he’s drunk he forgets to take his tablets”; and PLWAs, as well as alternative to clinics, such as treatment by herbs or prayer, or cures for AIDS (or “AIDS has no cure”). f) Alcohol contributes to getting AIDS (forget to take tablets, takes partners without forethought) , also use for alcohol leads to beating the wife, which is against the norms of marriage. g) Stigma h) Insults (for norms) Record Keeping: Unfortunately, there isn’t a function in Adobe that allows one to search for a code that’s been typed onto the scan (or at least neither I nor the Swaziland coder has found such a function). Since for the report we will need to make counts of conversation topics, I think it would help if the coders keep a record of each excerpt that was coded, along with identifying information. I suggest a Word document or an Excel spread sheet: Name of journal Page #s Codes Title of journal 100928-1010232 Pt 1. pp 1-3 1a, 4d Thus, if we're interested in whether this is comparing what people say about violence in marriage vs in casual relationships, we can count the proportion of the refs on violence that are violence in marriage and the proportion that are violence in casual relationships. Suggested coding protocols for coders Please add to or amend these protocols if there is agreement that they hinder effective coding. The important thing is that all coders understand, agree to, and use a set of protocols consistently. Use 2 coders to code each journal. Switch teams regularly. This will maximise debate and equalise any coder bias. Although this will significantly increase the time taken to code each journal, there is good evidence that this will preclude the need for inter-coder reliability checking – itself a very time-consuming process. It will keep coders engaged and minimise fatigue and memory errors. I feel that the extra time and cost will be justified by the improved quality of the data, especially given the time and cost already invested the project. Coders should take care to not get stuck on disagreements. In the case of two equally strong and opposite arguments, engage a third party as quickly as possible, accept the call, and move on. It is not an exact science, and no matter how much care is taken or time spent, there will always be a margin of error when interpreting second-language speakers’ recollections of conversations. Just be able to justify a decision. Don’t get too invested in an opinion. Remember that most codes are not mutually exclusive. Unless there is diametric opposition, use both codes. Different coders will have different insights, which enrich the data. Remember that this is the point of doing it as a team. In cases where local knowledge is important to fully understand a situation, please add notes to the code explaining a particular phrase, custom or convention of speech or behaviour that might not be familiar to every reader. Remember that further analysis might be undertaken on your coding, so do what you can to facilitate future scrutiny. 48 Review the object of the study, the codes and this document continuously. Five to ten minutes spent on this before every journal is not extravagant, especially in the first couple of weeks of coding. We are looking for anything that could viably be used by PSI to develop or modify an existing or future intervention or behaviour change programme or product. We want to understand how people think about relationships and HIV. We want to know what factors influence their behaviour. We want to know what people think and know about past and present HIV prevention campaign topics: Condoms, circumcision, multiple and concurrent partnerships, etc. We want to keep an eye out for avenues (openings) that PSI could in future use to deliver its messages and products. Common coding problems include the following: Irrelevant data It can become tempting to try to apply codes when reading pages and pages of irrelevant content. Resist the temptation, but read carefully. You might find code-able content in between loads of blather. Bad data You will come across spurious entries and conversations. If there is even some agreement between the coders that an entry is suspect, discard it by marking the original clearly and move on. It is better to have less data than bad data. You will often find these entries at the end of journals, as ethnographers rush to complete a journal before collection dates. First journals also sometimes contain made-up conversations, as the ethnographers have not gained confidence and seek to fill the journal with what they think we want to hear. Final journals also suffer as ethnographers are fatigued and have become more sophisticated in filling journals. You’ll notice that script often becomes larger and spacing more generous. Some will invent content. A lot of made-up conversations have much better language than previously exhibited, as ethnographers copy text from pamphlets and other publications. If a conversation seems stilted, too one-sided, peppered with too many facts or unnatural in any way, regard it with suspicion. The question-and- answer conversation occurs often. No-one talks like that, so watch out for those. Repetition People will often repeat themselves in the course of a conversation, in different entries in the same journal, or even across several journals. This presents a bias problem. I think the following is a fair solution: If the person presents the same argument against, say, circumcision several times in the same conversation, code for that only once. If they present different arguments against circumcision, code once for each of those reasons. If they present the same argument in different contexts, I would code for it with each new audience. The rationale is that we want to know how deep an opinion or attitude penetrates. The more it is expressed, all else being equal, the more effect we can assume it has, and the heavier it should be presented in the coding. Having said this, some journalists describe a regular cast of characters and some of these characters might feel strongly about a particular topic and issue opinions and arguments about it on a regular basis. In these (hopefully isolated) cases, I have previously coded heavier for such a person for the same reason stated above – I’m assuming that, all else being equal, the effect of this person’s opinions has 49 more of an effect on the knowledge and opinions of others than someone who might say the same thing only once. This is really a judgement call and I don’t want to offer any hard and fast rules for this kind of situation. Perspective You want to code from the perspective of the speaker. If you’re coding for knowledge, for instance, you will code for the speaker’s stated opinions, even if they are incorrect. You might think the speaker is coming from a position of ignorance, but that is about YOUR knowledge. The speaker believes they know so code for that. Coding vs. Analysis It might be tempting for you to over-think the coding exercise, considering that you have already come up with dozens of analytical categories. Remember that the point of the exercise is to put relevant content in “big piles” so that further analysis can be performed on it in a second stage. THE CODES H+ or H- Use these only when there is mention of HIV or AIDS. H+ anything about preventing HIV or solving the problem. Eg. “I’m getting circumcised because I don’t want to get sick” H- anything about a behaviour contributing to HIV. Eg. “He should stop proposing every girl when he’s drunk. He’s spreading this thing” A(pprove) or D(isapprove) A (of circumcision) “I’m getting circumcised because I don’t want to get sick” D (of casual sex/multiple partners) “He should stop proposing every girl when he’s drunk. He’s spreading this thing” MS (male speaker) or FS (female speaker) . This gives us an idea of the gender split in any value or opinion MA (about a male, or males in general) or FA (about female/s). This gives us an idea of what people are doing, what behaviours elicit comment, and so on. This could be used to target interventions. Openings - Use this code to highlight any venue, group or media that prevention programmes could viably use in a marketing exercise. Places where people gather, media people refer to, Examples include bars, clubs, shebeens, schools, taxi ranks, bus stops, Facebook, radio, TV, cell phones, billboards, organisations like Humana, radio/TV personalities, month-end or payday gatherings etc. Locus of control - Use this when speakers indicate anything about a behaviour, situation or outcome not being under their control, or more specifically, being under the control of another person, agency or drug. Eg. “I can’t use condoms, because then she’ll suspect me and leave me”; “I can’t say no to him – he’s paying the school fees”; “This place makes everyone a prostitute”. Be careful not to confuse this with motivation: “ I can’t help myself, I just love big bums” is not a comment on locus of control. “When I’m drunk, I can’t leave big-bottomed girls alone”, is. 50