This paper employs quantitative analysis to explore which people in Khayelitsha (an urban African community in Cape Town) are likely to be clients of ‘sangomas’, that is, traditional healers who specialise in divining illnesses usually perceived to be caused by witchcraft. It shows that sangoma clients are older, disproportionately female, poorer and less well educated than other
people and that they are less trusting of others and more likely to believe in the efficacy of witchcraft. Being a recipient of a disability grant is the most significant predictor of whether the respondent is a sangoma client or not. The paper also discusses different quantitative sources for the use of traditional healers in South Africa, showing that the way the question is posed is all important.
This paper employs quantitative analysis to explore which people in an urban African community are likely to be clients of ‘sangomas’, that is, traditional healers who specialise in divining illnesses usually perceived to be caused by witchcraft. It uses data from a survey of residents of Khayelitsha, Cape Town’s largest African township (home to over a third of the total African population in the metropolitan area).
The first houses in Khayelitsha were built in 1984. Since then, the township’s informal and formal housing areas expanded rapidly and by 2001, the population had reached 327,000. Khayelitsha’s residents include those who moved to the township from surrounding squatter camps and back-yard shacks when the first houses were built, and many people who migrated to the township (mainly from the Eastern Cape) over the past two decades. The population thus varies significantly in terms of the relative balance of exposure to rural and urban life. If the belief in witchcraft is more common in rural communities, this could affect the demand for sangoma services. However, as traditional healing paradigms have proved flexible in the face of urban living in other parts of Southern Africa (see, for example, LeBeau, 2003; Dillon-Malone, 1988: 1159-60), the degree of ‘rurality’ of social background may prove irrelevant. Indeed, if suspicion of witchcraft is a function of competition, insecurity and income inequality (Ashforth, 2001, 2005), then one would expect to find a significant portion of people in urban areas also manifesting witchcraft beliefs. Khayelitsha is a good place to test this and other possible socio-economic determinants of who becomes a sangoma client.
The quantitative analysis draws on a 2004 representative survey of 570 people in Khayelitsha conducted by researchers from the Centre for Social Science Research. This was the second wave of a panel study in which respondents had first been interviewed in 2000 as part of a wider survey of labour-market behaviour.1 Amongst other things, the Khayelitsha 2004 survey asked respondents if they had consulted a ‘sangoma’ the last time they were ‘very sick’. This paper uses this data to probe the characteristics of people who responded positively to this question and thus can be regarded as ‘sangoma clients’.
The Zulu word ‘sangoma’ (or isangoma) is conventionally used in African studies to describe traditional medical healers whose methods of diagnosis are linked to divination through communication with spiritual others, most notably ancestors (Henderson, 2005 – this volume; Wreford, 2005b, 2005c – this volume). Sangoma specialise in divining, healing and protecting against witchcraft (Good, 1987; Gelfand, 1967; Ashforth, 2005; Wreford 2005a). Sangomas appeal to their ancestors for help in diagnosing problems and
prescribing remedies and routinely check to see if illnesses are caused by their clients having violated cultural norms and traditions. Connections have been drawn in this respect between sangoma practice and the central role of the collective unconscious in Jungian psychotherapy (Buhrmann, 1984; Wreford, 2005b). However, unlike psychotherapy which ‘cures’ by facilitating selfunderstanding, sangomas ‘cure’ through cleansing rituals and by prescribing herbal and other remedies (mostly purgatives).
Thirty-five respondents – that is, 6.1% of the total – in the Khayelitsha survey said that they had consulted a sangoma the last time they were ‘very sick’. Of these sangoma clients, 19 (56%) said that they had visited more than one sangoma. The most common reported diagnosis by the sangoma was that the respondent had been bewitched or poisoned – often both. (Note that the word
poisoning in this context is indicative of witchcraft because “a notion of ‘poison’ serves as a basic building block for interpreting the dangers arising from the domain of ‘witchcraft’” (Ashforth, 2005: 144)). Of those who reported that the sangoma told them that they had been bewitched, one third reported that the sangoma had told them that they had been bewitched by their mother or a close family member. Just under a third of respondents reported that the sangoma had told them that they had angered their ancestors. In other words, the reported diagnosis by sangomas is consistent with what we know from the relevant
anthropological literature about the witchcraft paradigm within which they function.
With regard to the treatment strategies of sangomas, these appear to have followed the typical sangoma prescription of cleansing through purging – that is, instigating vomiting or diarrhoea (sometimes both), with an emphasis on vomiting (88% of sangoma clients reported being given a substance to make them vomit). Eighty-five percent of sangoma clients said that they had been given medicine to stop them being bewitched – and of these, most (90%), were also given purgatives. In addition to purging, sangoma clients reported being given remedies to ‘clean the blood’ (68% of sangoma clients) and make them ‘stronger’ (90% of sangoma clients). Most sangoma clients reported that they were fairly satisfied (although not enthusiastically so) with the treatment offered by the sangoma (20% said it helped a lot and 63% reported that it helped ‘a bit’).
This paper investigates whether there are any statistical regularities which could help predict whether a respondent is likely to be a sangoma client or not. Although 35 respondents constitutes only a small cell of individuals, the statistical analysis reported below reveals some interesting statistically significant patterns.2 Before doing so, however, it is worth placing this figure of 6.1% in a comparative context and exploring the limitations for the study that are posed by the way the question was asked.
See Magruder and Nattrass (2005) for a discussion of this panel study.
This analysis was conducted using Stata 8. The relevant do-file is called ‘sangoma.do’ (to be run on ‘okpanelkmp0004’ – that is, the panel data set comprising the 570 people from Khayelitsha who were interviewed in 2000 and 2004). The do-files and data set are available on request.