Researching Reproduction: Reflections on Qualitative Methodology in a Transforming Society

Abstract: In this article, the researcher considers ways in which qualitative methods could be used when engaging in research on reproductive and sexual practices. The primary method in South African demographic research is the sample survey, which has entrenched its status as a source of "reliable" and "scientific" data. The drive, in the post-apartheid context, for increasing quantities of credible data for policy and planning purposes has not created considerable space for discussion on the role of "softer" or qualitative approaches. Whilst qualitative studies do hold importance as ad hoc contributions, they are rarely considered by demographers studying women's fertility to be viable alternatives to large-scale survey research. As South Africa braces itself for higher levels of mortality due to AIDS-related deaths, qualitative methods are being utilised to build subjective understandings of the AIDS-fertility relationship but rarely in terms of exclusively qualitative research designs. The article reflects on longitudinal fieldwork and focuses on alternative and "mixed" approaches in which qualitative methods could be drawn upon to illuminate the various facets of women's personal, social and cultural existences.

Key words: fertility, qualitative methodology, mixed methods, Winterveld, South Africa

3. Historical Legacies and Particular Difficulties with Survey Data in South Africa

4. Reflecting on Difficulties in the Field: Misunderstandings in a Socio-cultural Context

5.1 Case study 1: AXINN, FRICKE and THORNTON's (1991) micro-demographic community study approach

5.2 Case study 2: BLEDSOE, BANJA and HILL's (1998) open-ended survey and investigative fieldwork approach

5.3 Case study 3: Nancy SCHEPER-HUGHES's (1997) critical interpretive demographic approach

How do demographers interpret responses and make claims about, what are categorised as, "the proximate determinants of fertility"? The researcher reflects on continuing work in a particular South African fieldwork site in which an attempt was made, eight years ago, to conduct a micro-survey modelled on the standard demographic and health survey to probe women's childbearing, fertility preferences and "reproductive decision-making". This project was reassessed as the social environment presented itself as too complex for a simple survey and as the questions answered begged further questions and as the research design shifted from being solely quantitative to almost solely qualitative. South Africa, as is the case with many transforming societies, faces enormous challenges with regard to building substantive understandings of the various forces driving and debilitating its fertility transition. Critical questions are being asked about HIV/AIDS and how fears of infection might be serving to shape reproductive dynamics in different localities and the society as a whole. And, at the same time, there is much lamenting that little is known about how fertility will be shaped in the future. My purpose in this paper is to consider some of the problems posed by survey data and then outline ways in which qualitative methodology can make a contribution towards building more in-depth and dynamic conceptions of the issues shaping reproduction and family life. [1]

Demographers studying reproductive behaviour and fertility transition have, over the years, depended almost entirely on data gathered from sample surveys, censuses and registration systems. Surveys maintain a central role in demographic research and make an important contribution towards the construction of reliable, scientific and large-scale national and regional understandings of populations. In the past few decades survey programmes, such as the World Fertility Survey (WFS) and Demographic and Health Survey (DHS), have gathered increasing quantities of data on aspects of reproductive behaviour and on fertility levels and trends in most developing countries. Prior to these interventions a paucity of data had existed. With greater efficiency in organisation and with better-trained staff the organisers of the WFS and DHS programmes have begun to facilitate not just the collection of data for individual countries, but the comparison of trends across countries and continents (UN, 1995). Follow-up DHSs in individual developing countries and regions have been designed and undertaken to assess changes in the proximate determinants of fertility and to answer questions on whether or not these countries are undergoing a process of fertility decline. [2]

Alongside the growth in fertility data and some praise for the achievements of the survey programmes has been increasing questioning of the quality, purpose and meaningfulness of the data gathered through surveys. There are many reasons why surveys are frequently criticised. Some of these reasons include: (1) problems associated with the standardisation of questions asked irrespective of global region and limited inclusion of new "locally relevant" questions to probe various phenomena and (2) the inability of surveys to illuminate dynamic processes, but rather snapshots of demographic trends (CALDWELL, 1985); (3) the great dependence on hired interviewers who are entrusted with the huge responsibility (in the case of DHSs) of extracting from women (whom they do not know) intimate details of their marital, sexual and reproductive lives within a short space of time. In addition, (4) fertility survey programmes such as the WFS and DHS have also been questioned because, despite claims of political neutrality, they serve the particular political agendas of governments who seek to influence reproductive trends (BROWN, 1987). In this regard, gender activists have raised concerns about the targeting of individual women as objects to be controlled and manipulated (KABEER, 1994, p.187). [3]

The inherent limitations of surveys have since the 1980s been a subject of much discussion and debate and have led to experimentation with alternative methodologies. The pioneering efforts of John and Pat CALDWELL in making use of ethnographic methods in demography raised initial arguments on the need to go beyond snap-shot surveys and to develop instead, through extended periods of fieldwork, more in-depth analyses of the socio-cultural contexts and processes shaping kinship relations and reproduction. Although John CALDWELL wrote his critique of fertility surveys in 1985 it was only perhaps since the early 1990s that discussion of the integration and use of qualitative techniques and research designs became a lot more vigorous and comprehensive. Perhaps three noteworthy concerns have arisen in recent symposiums and forums focusing on methodology and demographic research. These are concerns about:

3. Historical Legacies and Particular Difficulties with Survey Data in South Africa

The politics of the early demographic fraternity in South Africa stands in direct contrast to the values espoused by a "critical demography". Rather than a critical and engaging discipline demography found itself aligned to the conservative policy-making agenda of the South African state through the articulation of growing concerns about rapidly increasing population numbers and about the political and economic implications of high fertility amongst the black sector of the population. Alarmist neo-Malthusian projections of an impending population explosion were developed which fed into fears that black population numbers were going to increase at an uncontrollable rate (SWARTZ, 2003; BROWN, 1987). Thus, family planning which had mainly catered for an urbanised, English-speaking sector of the white population became a project for black women by the 1970s. BROWN (1987, p.264) maintains that: "once the government decided to launch a programme to control women's fertility, it spared no expense". Having struggled for government support to strengthen and spread its services to black areas, the voluntary sector was overwhelmed by a massive injection of funding and aid in the period between the late 1950s to the early 1960s (CALDWELL & CALDWELL, 1993, pp.22-24), and was eventually overtaken by the more organised government initiative in the 1970s. From the 1970s to the early 1990s, family-planning programmes were integral to the provision of general health services. The Population Development Programme (PDP), established in 1984, mobilised a wide spectrum of actors into the national initiative: educators, social workers, medical practitioners, and media workers were all used to legitimise the idea that benefits were to be derived from a reduced family size. Most attention was paid to contraceptive provision in urban (or peri-urban) areas, although rural areas were also catered for, through mobile services (CALDWELL & CALDWELL, 1993). The PDP also set itself targets. For example, it aimed to influence a decline in the fertility rate to 2.1 children per woman by 2010 (NDEGWA, 1996). The PDP's main difficulty was that of political legitimacy. As NDEGWA (1996, p.54) suggests:

"The fact that it centred on a community with no vote and therefore had no say about its existence, coupled with its participation in the National Security Management System, which was set up to undermine community mobilisation against apartheid and specifically the UDF [United Democratic Front], made it look, in the eyes of the majority population, like an apartheid government tool to reduce the black population." [5]

However, despite the climate of suspicion and condemnation of contraceptive use in the 1970s and 1980s, black women, including women in the Bantustans¹⁾, began to make use of contraceptive services in increasing numbers (KAUFMAN, 1998) seeing the control of fertility in terms of strategies to "free themselves from traditional patterns of marriage and patriarchy" (MOULTRIE & TIMAEUS, 2001, p.223). CALDWELL and CALDWELL's (1993) assessment of the old South African Family Planning Programme was that it was not simply a strong one, but an experiment of global significance. The South African case, they suggest, reveals ways in which a highly efficient programme can influence changes in fertility, notwithstanding adverse socio-political and economic conditions. [6]

With transition to a post-apartheid society and with the discourses changing from the need to influence fertility reduction to the need to enhance women's empowerment and improved health a new generation of fertility specialists find themselves with new challenges (ANDERSON, 2003). Whilst it is common to hear strong criticism of the research of past eras and read arguments that the contextual and dynamic realities influencing fertility behaviour should be interrogated more rigorously, in practice demographic methodologies remain largely unchanged with the key priorities of demographers remaining to monitor changes in fertility, reproductive practices and reproductive health and to serve as informants of a changed (albeit a democratic) government's policy-making institutions. Sample surveys remain the primary method for gathering data, and questions relating to "methodological mixes" are virtually absent in contemporary discussions on fertility research. Policy-makers intent on developing reliable facts have prioritised the search for improved macro-level indications of changing fertility to help the policy processes. [7]

CALDWELL's introductory comments in The South African Fertility Decline summarise, to some extent, academic opinion on the state of fertility data in South Africa prior to the mid-1990s. He suggested that for a variety of reasons data on the black population had been generally "too deficient to be useable" (CALDWELL & CALDWELL,1993, p.225) but that more satisfactory data became available after the results of the 1987-1989 DHS-type survey were published. Although he notes that there were many unanswered questions on this survey's methodology, a volume of important fertility data became available. More data has become available through the 1998 and 2003 demographic and health surveys. The 1998 DHS was launched under the auspices of the Department of Health, South Africa's national state health institution, and has been to date a widely used resource in analysis of fertility change in the current era. This DHS, which has had as a key objective the need to "improve demographic data in South Africa", benefited from very wide consultation with a spectrum of interest groups and international specialists. It has been heralded as South Africa's first true demographic and health survey and has been based on the standard international DHS format. It has integrated a wider range of questions, to address specific South African policy needs in health and welfare, than commonly seen in demographic and health surveys. There were questions on the chronic health problems of women and knowledge and awareness of HIV/Aids (XIPHU, 1998). The broader objective of the survey was to draw on existing expertise and to "build research capacity of the organisations involved, e.g. provincial departments of health ... in planning and implementing a survey ..." (KLUGMAN, 1996, p.5). This capacity was further enhanced in preparation for the 2003 survey. Against the background of the drive for more sophisticated data analysis, ultimately for the use of transformative state interventions in the health and welfare field, qualitative data seem to emerge as poorer and second best. In the South African context, surveys are, nonetheless, far from problem-free. Some difficulties with the measurement of socio-economic differentials (ethnicity, rural-urban divides, work and education) in demographic and health surveys are highlighted below. [8]

The problems associated with using quantitative data are most evident when attempting to aggregate and average socio-economic differentials. The first crudity is one which KLUGMAN (1996, p.3) suggests is a "positive decision"—that is, to continue to disaggregate data in terms of "the old "population groups", since they have served as proxies for socio-economic status". She adds that it is essential to retain these "so that changes in key indicators of quality of life can be measured." Whilst there was some justification for doing so in the mid-1990s, they are not so apparent in the mid-2000s. Race appears to have become reified in survey research and distanced from historical studies and debates on the problematics and politics of simple racial categorisation. With regard to the association between ethnic categories and reproduction, JAMES and KAUFMAN (1997, p.1) maintain:

"we argue that this association derives meaning from the process of ethnic identification and the cultural and historical construction of that identity." …

"Given a strong association between ethnicity and reproduction beyond, or informed by, social and economic status, demographic analysis must address those sets of behaviours, attitude, or contextual factors accompanying ethnic identification that might give rise to particular reproductive patterns". [9]

For example, the 2003 demographic and health survey shows (for the first time) that the total fertility rate for "coloureds"²⁾ (2.3) is now higher than that of "Africans" (2.1) without any explanation for what contextual, behavioural or related factors has given rise to this recent transition. The higher "coloured" TFR is intriguing and unanticipated given that, as a group, "coloureds" are financially better-off, more urbanised and have access to better health services, than the African sector of the population. Another key differentiating feature in demographic surveys is urban-rural residence. [10]

Although there is recognition of the overlapping between "rural" and "urban" and the various categories that fall in between, the new surveys continue to place emphasis on contrasting "urban" and "rural" (or non-urban) behaviour. In analysing models using rural-urban concepts, HARRISS and MOORE (1984, p.24) suggest, "the rural-urban dichotomy has been asked to bear a heavy burden. In terms of economic activities there is often more overlap between and differentiation within the two sectors than the theorists' models would imply". This burden is particularly problematic in the South African case. There has been, over the years, considerable movement from rural to urban areas due to dispossession of land, the migrant labour system and the search for employment. The consequence of displaced urbanisation (MURRAY, 1987), movement between rural villages and urban centres, and the growth of peri-urban areas and squatter settlements on the fringes of towns is that a sustained linkage between "rural" and "urban" exists. With the linkage emphasised, is it sensible to talk in terms of discrete urban and rural influences on, for instance, fertility behaviour? South African survey results have tended to emphasise that urban and rural differences are real: differences can be discerned in contraceptive use and desired fertility. Sometimes tensions are unsurprisingly apparent. For example, in analysing some results of the 1987-1989 DHS, MOSTERT (1990, p.66) offers general points about the projecting of urban-like influences (with respect to contraceptive use) on to the people "within its vicinity". The analysis, however, becomes difficult when he reveals that the Pretoria-Witwatersrand-Vereeniging (PWV) district displayed some of the lowest levels of contraceptive use. He explains this by acknowledging a mixing of influences: he noted that the PWV, as an urban area, has "an influx of people, probably mainly from the rural areas" (MOSTERT, 1990, p.66). The data are also beginning to show less evidence of difference between rural and urban. The 1987-1989 DHS (DU PLESSIS, 1996) and the 1993-1994 DHS (NKAU, 1999, p.4) show no significant differences in desired fertility among younger women living in areas categorised as rural or urban. The 2003 SADHS shows that the TFR in both rural (2.1) and urban (2.0) areas are about the same. In the highly mobile peri-urban spreads or interfaces around the country, most people would not be easily placed in "rural" or "urban" categories. [11]

Similarly, the deep insecurities surrounding women's work and changeability of occupational role undermines neat associations between work and fertility. Survey evidence shows that although working women give birth to and desire fewer children than unemployed women, the differences are minimal. The need to know (through access to better-quality data) about "whether specific kinds of employment make differential impacts" (KLUGMAN, 1996, p.5) demands a lot from micro-level quantitative data. Perhaps it would be more reasonable to begin to answer such questions through qualitative interventions in messier social realities—realities in which not just the specific kinds of work can be assessed in order to learn about differential impacts, but also the complexity of work experiences in shaping women's lives. It may not be just the kinds of work that women do which are important to examine, but also fears of job losses, the wages earned, the distribution of household income from which women may or may not benefit, and the various ways in which women might sustain livelihoods. All these have implications for reproductive decision-making and well-being. The 1998 DHS shows that high proportions of South African women are "unemployed" but does not interrogate domestic work, partner's work or informal sector work as factors influencing fertility. [12]

Surveys have been a lot more successful in measuring associations between educational levels and reduced desires for fertility. The positive implications of increased investment in schooling have been emphasised quite strongly over the years. Both the 1987-1989 and 1998 SADHS show that high school education plays an important role in discouraging high fertility over the reproductive life span. The post-apartheid South African state has sought to address gender inequalities in access to schooling and to improve the quality of girls' education. It is anticipated that the long-term effects of continuous exposure to formal education will act to empower and enhance the position of women generally. Schoolgirl dropout rates however are relatively high and the "problem" is often treated (by state officials) as one that can be studied, explained and remedied through improved research and interventions in the school and the family. Redressing the problem, however, requires more than "reliable" statistics on how many teenage girls dropped out of school because of an unplanned pregnancy. It will require textured analyses of teenage experiences. High levels of teenage pregnancy indicate many things surveys are unable to reveal. There are various motivations for early sexual activity. Many of these motivations are not easily established and a careful and deliberate attempt to understand the issues around teenage pregnancy could reveal masses of "data" on the contexts in which reproduction occurs, and what interventions are possible. [13]

4. Reflecting on Difficulties in the Field: Misunderstandings in a Socio-cultural Context

The foregoing has been important to set the scene for a personalised account of an experiment with survey work in a South African field site. In 1998 I set myself the challenging task of conducting a micro-survey modelled on the DHS format to probe the proximate determinants of fertility in the Winterveld, a South African locale inhabited by about 150 000 people, that could fit the description of both rural and urban. A random sample of 293 women within a central demarcated part of Winterveld, between Jakkalas and Vezaubuha, was drawn. The site itself encompassed about 10,000 dwellings. Fieldworkers from the local Planned Parenthood Association assisted with the survey. The follow-up to this survey has been in-depth interviews and focus groups conducted, intermittently over the past nine years, by myself and a multi-lingual woman resident from the area (NAIDOO, 2002). [14]

The Winterveld is not by any description a unique South African place. It was established as a black agricultural holding settlement in the 1930s but became a densely populated multi-ethnic area in the 1980s when the apartheid state was relentlessly pursuing its strategies of social, political and geographical engineering³⁾. By the mid-1980s the area was inhabited by a mix of socio-economic classes but with a disproportionately high number of very poor people, many of whom were forcibly relocated or evicted from surrounding areas. People in the area waged many struggles for services and political emancipation in that period and continue to do so in post-apartheid South Africa. While the politics has changed there remains in the 2000s little infrastructure, predominantly self-built homes, and few basic municipal services. Despite this, people choose to live here because Winterveld offers the occasional work opportunity: it is 40kms north of Pretoria and on the taxi route to Johannesburg (90 kms away). The majority of people are unemployed, and domestic violence, stress and insecurity are pervasive (SIMONE, 1998). In these terms the area is not fundamentally different from many other parts of South Africa that display tensions, mobility, and instability of family life. How does one make surveys work in such terrains and how can researchers be sure of what is being measured and of what people report as being "true" of their situation? [15]

The proximate determinants of fertility (viz. entry into marriage, contraceptive use, abortion and breastfeeding) are useful starting points for critical reflection. The socio-economic variables, referred to earlier, are regarded as non-proximate variables and are seen to work through a set of (direct) proximate determinants to influence fertility (BONGAARTS, 1978). Marriage has an important place in the BONGAARTS model as an indicator of exposure to sexual intercourse and risk of pregnancy. Thus, measuring the proportion married, the ages when first married and sexually active, allows for estimates of children being born and children likely to be born in the future. Data on contraceptive use, induced abortion, breast-feeding (as post-partum sterility) and fecundity (as influenced by STDs) reveal important information on fertility regimes and factors controlling, inhibiting and determining fertility outcomes. [16]

The confidence of demographers seeking to estimate marriage rates, or to measure the married sector, has waned over time. Whilst close to 50% of women aged 30-34 years old had their first child before 20 years (according to the 1998 DHS) most of these births were outside of marriage. MOSTERT (1990, p.67) in recognising the early pattern of births outside marriage claimed that, notwithstanding this situation, by the age of 50 virtually all (93.1%) will be married. MOSTERT (1990, p.68) concedes, though, that "[m]arriage dissolution through divorce is however high in all age categories—averaging about 15% in every category. The divorce rate has been at a high level for the past three decades." All South African DHSs, including the latest 2003 DHS, asked questions in an attempt to establish whether women are in "marital" or "living-together" unions, whether they are in polygynous unions, or whether women have been married more than once. Increasingly, though, South African researchers have become aware of the small difference marital status makes as evidenced by data on the lifetime fertility of women (between ages 45-49) who are married (4.34) and those who are not married (4.03) (NTSALUBA, 1998, p.39). [17]

In asking questions on marriage in my Winterveld survey it became clear that women might sometimes identify themselves as "married" when they were not in a conventionally defined "marital union". Instead, they might have been in a precarious union in which they saw their partner(s) infrequently. In such cases, a woman who was "never-married" might have the same frequency of contact with a male partner and would be likely to have similar needs for contraception as would a "married" woman. In addition, through the course of qualitative work, it also became evident that a large number of women were involved, simultaneously in numerous unions. Whether pursued for reasons of poverty or not, an involvement in multiple unions raises questions on contraceptive use, and on motivations for the timing of and desire for children. Surveys do not adequately capture the ways in which women contend with numerous and unstable unions. If surveys were to refine questionnaire items and include questions to establish the number of partnerships women might have and frequency of sexual relations (which some surveys, including the DHS, have attempted to capture), I do not believe that a one-off survey in which an interviewer attempts to speedily extract data from a respondent would be very revealing. In the 2003 DHS the following question is asked: "Have you been married or lived with a man only once or more than once?" Apart from the confusion that could stem from attempts to interpret women's responses to such questions, what does a woman achieve by telling an unknown researcher intimate and personal details? A question was asked in the 1998 DHS: "In the last 12 months, with how many different men have you had sexual intercourse?" Unsurprisingly, an insignificant number of women responded in the affirmative. Clearly, it is only after a long period of involvement in the lives of women that such information can be shared and clarified. [18]

When compared with levels of contraceptive use in other African countries, contraceptive use of African women in South Africa is regarded as being very high. The 2003 DHS suggests that 65% of sexually active women use contraception. This is not very different from initial admissions in my fieldwork site where approximately 70% of women of reproductive age claimed to be using contraception. Whilst previous programmes targeted women only, the spread of HIV has led to the establishment of health and reproductive programmes that require the active involvement of men. Health surveys have only recently begun to ask probing questions about men's use of condoms and about knowledge of AIDS. In the era of AIDS the 1990s became a challenging one for those involved in the construction of health surveys and their specific items. They were required to display sufficient understanding of cultural constructions of disease, health and ways in which and purposes for which contraception was used. In surveys in the 1990s questions concerning knowledge of AIDS and whether men were using or would be happy to use condoms, were asked without addressing related complications such as whether men would use condoms when they were with their wives or only when they were with other women? There are a number of questions that need to be asked about survey findings on contraceptive use. For example: do women who admit to using contraception also admit that they often use it inconsistently? Why is it that the 2003 data suggest that "there has been a decline in the level of knowledge of contraceptive methods since 1998"? (Department of Health, 2003, p.10). Why are women in the field less concerned than researchers about "unmet needs?" Is it partly because of some reliance on "folk methods" which researchers are unable to capture in survey data? Only 13% of women ever used folk methods according to the 1998 SADHS with even fewer women referring to them in 2003 (8.9%). In Winterveld numerous folk methods referred to as "cultural" methods are in use. The most common is burying bottled menstruation (mixed with herbs) outside your home. The bottle is only retrieved and opened when a pregnancy is desired. Despite claims that it is declining in importance, it has become increasingly important to understand the role played by traditional medicine in reproductive health. I learnt, through conducting my own survey, that in a snap-shot survey women tend to feel embarrassed to talk about using "cultural methods" as contraception particularly when interrogated by "well educated interviewers". Thus the tendency is to inflate the significance of modern methods in relation to others. [19]

In the course of fieldwork I have found that women are also shy to talk about their knowledge and use of traditional methods unless they are made to feel comfortable in doing so. Numerous demographers term folk methods (such as abstinence, withdrawal and rhythm) as "inefficient methods" (MOSTERT, 1990; DU PLESSIS, 1996). Clearly, they are less effective in preventing pregnancy than the pill or injection. But when the pill or injection are used inconsistently or are given up in favour of traditional methods, then these modern methods also become inefficient. Surveys have not adequately captured the extent to which traditional methods are used solely and in combination with modern methods. This is partly because, when being interviewed about contraception, women prefer to reveal use of more modern forms of "prevention". They might, for example, not consider abstinence as a "method" and therefore not disclose it; interviewers are also unlikely to take time to probe periods of deliberately practised sexual abstinence. Withdrawal is even more difficult to learn about. It is a method dependent on the co-operation of men; women might be reluctant to mention it as a method which they use. I have found it to be widely practised in Winterveld particularly in the post-partum period when sexual relations are often taboo. Withdrawal is also practised when men and women engage in infidelities. [20]

It is of course difficult to learn about abortion in any setting. If it is difficult to construct a picture of how prevalent the practice is through long-term ethnographic work, it will be impossible to ascertain anything insightful through surveys. The 1987-1989 SADHS did not ask any questions about it because abortion was previously illegal. Illegality has rarely prevented women, in particular schoolgirls, from resorting to an abortion if the need for it arose. Arguably, it is not the fact that it was previously illegal that makes women hesitate to want to talk about abortion; secrecy is usually maintained because of the fear of negative cultural sanctions. At the same time, although abortions are frowned upon, virtually everyone I interviewed (in my more in-depth interviews) reported that they knew that these were being frequently performed. In the survey not many women disclosed whether they had had an abortion or whether they would resort to one if faced with an unwanted pregnancy. For the same reasons, the 1998 SADHS did not probe or ask women whether or not they had had an abortion. The survey was able to test knowledge: slightly more than 50% of all women were aware that the law had changed and that it was legally permissible to get an abortion. It is important for fertility analysts to establish just how many women continue to end pregnancies, both in clinics and in backstreet operations, rather than give birth to unwanted children. Surveys, however, are not useful instruments for eliciting such information in the South African context. [21]

South African surveys are beginning to gather useful data on breast-feeding and post-partum sexual abstinence. Questions however are not being directed towards ascertaining women's knowledge of the role of breast-feeding in lengthening birth intervals. No questions have been asked to date on amenorrhoea and its association with "full" or limited breast-feeding. The 1987-1989 DHS relied on establishing "for how many months did you breast-feed (the last child)?" There are obvious problems with memory. The 1998 and 2003 DHSs ask the same question but attempt to remedy limitations here by looking at supplementation of breast-feeding with other kinds of feeding, such as bottle-feeding, but do not probe whether mothers use breast-feeding to extend birth intervals. [22]

In the course of fieldwork in Winterveld, I found that questions on duration of breast-feeding asked on different occasions provided different answers. I was often surprised by the unhesitant way in which women of all ages would initially offer precise answers to the question of how long they had breast-fed their last child. When probed, women often revealed that they had relied on other indicators, e.g. they remembered that when the child in question had started to walk, talk or celebrate a birthday they had either just stopped breast-feeding or were about to stop breast-feeding. These indicators can hardly result in precise estimations. All the surveys show breast-feeding durations to be fairly short in South Africa when compared with that in other African countries. In Winterveld I found that women tend to present breast-feeding duration as being shorter than it was in practice in cases where the post-partum period of sexual abstinence had been reduced. In other words, it is embarrassing to disclose that you have continued to breast-feed whilst engaging in sexual relations. In Winterveld, at least, there remains a belief amongst a fair number of women about the harm that is caused to children if breast-feeding is not accompanied by sexual abstinence. Whilst all women do not subscribe to this belief, most would be reluctant to admit that they have been breaking cultural rules. Hence it is possible that the length of breast-feeding revealed by surveys is deliberately underestimated given cultural beliefs and social expectations. [23]

A purpose of this paper has been to draw attention to some of the difficulties surveys regularly face in attempting to build "accurate" representations of what influences and determines fertility behaviour in South Africa. In some countries, e.g. Uganda (BLANC et al., 1996), attempts have been made to overcome some of the difficulties of surveys by sensitising questionnaires to cultural and gender dynamics and to link them to focus groups to allow for probing of how reproductive decisions are made and how they change. In a period in which South African policy-makers are mobilising greater expertise to gather more useable data, not many questions are being asked about the inherent difficulties of surveys as research instruments. Greater care in questionnaire construction and in the conducting of fieldwork can play an important part in withstanding some of the problems posed by a complex and fluid South African social context. Despite isolated calls for a greater multi-disciplinary involvement in demography (DU PLESSIS, 1995) to reshape its tools and underlying philosophy, fertility research remains primarily quantitative. Anthropologists, historians and sociologists using qualitative methods have established an extensive literature on South African family life and kinship over the years. The extent to which these contributions can invite collaborations and methodological experiments requires greater investigation. There are three possibilities for experiments that I refer to below. [24]

The international literature is now abundant with case studies using qualitative methods for demographic research. I cite three exceptional case studies, which, although initiated in the 1990s, offer useful lessons for application both in the contemporary Winterveld and a broader South African context. All three are studies that have been conducted in poorer, developing social environments making them especially relevant to the current study. [25]

5.1 Case study 1: AXINN, FRICKE and THORNTON's (1991) micro-demographic community study approach

In AXINN et al.'s study of social change, family process and fertility among the Tamang in Nepal the attempt was to simultaneously apply survey and ethnographic methods within two sites. A sample survey was not conducted. Every person over the age of eleven (a total of 1,415 people) in two small communities was interviewed. The researchers who initiated the study called themselves the investigators. It was the task of the investigators to become familiar with the cultural context and to get to know the language, the histories and social activities of the local people. It was only after this initial process that the research instruments could be meaningfully structured. These included the household census, family genealogy schedules and an individual questionnaire. For the purpose of administering these formal instruments interviewers from the local community were recruited and trained. Survey work was undertaken immediately after investigators had visited the individual household as part of the introductory and rapport-building process. Investigators and interviewers lived together in common quarters. Each evening survey interviews would be cross-checked against ethnographic work for errors and omissions. In addition to reducing non-sampling errors, this integrated approach was useful to the investigators because it raised awareness about issues and events that were not considered prior to the survey being constructed. It thus allowed for the adaptation of the questionnaire in the field. New categories on clan membership, marriage, and expected behaviour in families were included. AXINN et al. (1991, p.212) maintain, thus, that "[q]ualitative data-gathering techniques designed as a simple supplement to surveys, which focus on the topics covered in the survey, are not likely to provide the means of attaining ... additional goals." [26]

In combining rigour in the application of survey methods with fieldwork efforts to become sensitised to the historical and social context within which the Tamang people live and interact, AXINN et al.'s work has gained much credibility. My survey in Winterveld preceded a qualitative investigation of women's lives and would have benefited from a more coherent connectedness between the micro-survey and the fieldwork. My purpose in conducting the survey was partly to develop a sense of what patterns existed with respect to particular aspects of fertility behaviour and partly to assess how the patterns in this area differed from those evident in large-scale national surveys. Some important differences were discerned such as the larger number of women living in informal unions, higher levels of contraceptive use and less evidence of the role of education in shaping fertility preferences. The experience of conducting the survey, however, became more useful for raising problems associated with the asking of sensitive questions in a brisk manner. An application of AXINN et al.'s (1991) methodology could make surveys a lot more workable in Winterveld. In saying this, however, there are limitations to using this approach in Winterveld or similar kinds of South African environments. AXINN et al. (1991) depict the Tamang communities as though they are both stable and "captive audiences". There is relative uniformity in custom and social practice, enabling a problem-free data-gathering process. Investigators and interviewers will find that they are less capable of such rigour in many South African areas. Identities, practices and family formations are diverse. Instability is a feature of everyday life to the extent that an entire "community" will not be easily "contained" as an object of study by an outside group. A large consignment of researchers who root themselves in the area for a comparatively short space of time is bound to be resisted, manipulated and tested for, amongst other things, financial benefits by those being researched. I am also doubtful whether reliance on insider-interviewers to elicit detail on intimate subjects such as sexual behaviour and reproductive practices will gain the most insightful results. [27]

5.2 Case study 2: BLEDSOE, BANJA and HILL's (1998) open-ended survey and investigative fieldwork approach⁴⁾

In 1992 as the first phase of a study on fertility in rural Gambia, BLEDSOE et al. undertook a survey of 2,980 women in forty villages. In the course of the survey open-ended questions led to the gathering of extensive fieldnotes on certain questions, particularly with respect to contraceptive use and child-spacing. The findings revealed low levels of modern contraceptive use (mostly Depo-Provera) and low levels of traditional methods. It was an area in which high fertility was said to be valued. Here women sought to "gain a competitive edge over present and future co-wives and sisters-in-law by bearing a number of children" (BLEDSOE et al., 1998, p.22). Birth intervals were found to be regular (children were spaced about 2.5 years apart); the spacing of births had been analysed previously as being governed largely by natural fertility patterns. A closer and more qualitative look at the survey results, however, revealed other things not easy to explain. Among them was the need to explain why, out of the 150 women using modern contraception, 27 (18%) were using contraception after a miscarriage or still-birth (a "reproductive mishap"). BLEDSOE et al. stated that "[a]n outsider's first reaction might be to attribute these reported actions to data error or statistical aberration". The inconsistency as they saw it was threefold:

They then attempted to further investigate why women who had experienced recent reproductive mishaps, might want to use contraception and delay future births rather than attempt another pregnancy soon. In the course of their investigation an alternative view of reproduction and ageing was developed—one in which intentions to use contraception were linked with "health reasons" or the need to find "recuperative space" rather than for the reasons of birth-spacing or the cessation of child-bearing. In focusing on the meanings attached to miscarriages and still-births, BLEDSOE et al. (1998, p.49) raise criticisms of contemporary demographic approaches which treat "live births as the only reproductive currency". [29]

My purpose here is not to detail the findings of BLEDSOE et al. but merely to emphasise that an examination of issues which do not appear to "make quantitative sense" and that might seem contradictory to general patterns, can lead to important insights and alternative ways of looking at motives influencing reproduction. BLEDSOE et al. sought to explore anomalies presented by the quantitative data and to unmask complex motivations by studying responses to open-ended questions. There is surely sufficient evidence of anomalies in South African data to warrant a qualitative probing. Numbers offered by different datasets could be questioned with more critical insights drawn from investigative qualitative work. Perhaps an emphasis on small-scale studies to address gaps in existing data could be of greater value to the policy-making process than an onslaught of surveys and aggregate level statistics. [30]

The Winterveld case study has drawn some parallels in that it has also revealed varying motivations for contraceptive use which appear to be out of place but which are ultimately guided by a wide set of responses to experiences of still-births, disease and violence. It is not just survey work that reveals inconsistencies. Different kinds of qualitative work often reveal experiences that appear to be aberrant. With a mingling of people originating from all parts of the country and with the arrival of increasing numbers of Mozambicans and Zimbabweans in Winterveld, uncovering how fertility trends are being determined there will become an increasingly challenging task. Qualitative work could play an important role in examining seemingly contradictory factors and forces influencing fertility change. [31]

5.3 Case study 3: Nancy SCHEPER-HUGHES's (1997) critical interpretive demographic approach

SCHEPER-HUGHES's methodology contrasts quite strongly with that of AXINN et al.'s and BLEDSOE et al.'s. Her starting-point is that objectivist and interpretive frameworks offer opposing convictions about the nature of reality and that these convictions shape conceptions of what are seen as "useful data" and what form research should take. She rejects conventional objectivist, quantitative research in demography and suggests that whilst such research efforts

"can strive to be culturally sensitive and can illuminate, for example, the cultural logic and alternative shapes of rationality that may govern the fertility and reproductive decision-making of Third World women ... there is often a striking lack of awareness of the ways in which the culture of their science structures the questions asked and overdetermines the findings" (SCHEPER-HUGHES, 1997, p.203). [32]

She presents her critical interpretive approach as a much more radical undertaking. Her approach is a simple but intense one: it involves years of immersion within a fieldwork site in which the researcher is not a neutral recorder of facts but an active campaigner in the lives of people. Rather than being merely an observer the researcher is central to social interventions in the area. SCHEPER-HUGHES speaks with the experience of having remained within a particular fieldwork site (albeit intermittently) for 25 years. Her fieldwork in the Brazilian shantytown of Alto do Cruzeiro entailed a careful studying of the causes, meanings and effects of child mortality in a largely poverty-stricken area. In addition to drawing out reproductive histories, details about women's lives, their marriages, and their attitudes towards the deaths of their children, SCHEPER-HUGHES wrestles with moral dilemmas and with her status as an academic researcher. [33]

With the current emphasis of research institutes on short-term research for speedy results, few academics can afford 25 years for fieldwork. The advantage though of sustaining links with a particular small setting is that it allows for the deeper exploration of that which is hidden from public records and scrutiny. SCHEPER-HUGHES uncovered and spent much time exploring considerable underreporting of infant deaths. In Winterveld where secrecy has marked the nature of living over the years, infant births and deaths, HIV-deaths, marriages, separations and ways in which people earn a living are often hidden from official records. My period of fieldwork has now entered its ninth year and I am still uncovering new strategies of a changing population in which many people continue to live without birth certificates, identity documents and licences and where there is little interest in being accurate about numbers or about reporting births and deaths. [34]

My attempt was to engage a small group of "marginal" women in Winterveld, to hear their stories and, in a sense, to make their experiences more visible. Involvement in their lives, a practice unanticipated at the beginning of fieldwork, became imperative in time. It has not, as yet, become anything close to "a radical undertaking". Winterveld, as most other areas in South Africa, has not been short of activist campaigners over the years. Many of these campaigners are now part of the policy-making machinery of the post-apartheid democratic government. Rather than become a political campaigner, I aligned myself with achievable goals by linking some income-generation activities with fieldwork. Such involvement amidst the poverty in Winterveld did much to build relationships with women. It has also, for me, reinforced much faith in the value of long-term, "engaged" qualitative work (in particular, ethnographic work) for developing deeper insights into how poverty destabilises and keeps marital relations, fertility, family and "community life" in a state of flux. As general points, some qualitative work entails simply the analysis of open-ended questions asked in the surveys. Some may involve a small number of individual or group interviews detailing aspects of women's life experiences and reproductive histories. The strength of the analysis will depend on the opportunity taken by the researcher(s) to learn about the women being interviewed, and to become familiar with their social environment and experiences. A temporal perspective is thus crucial to the quality of a qualitative design. [35]

With respect to collaborative kinds of involvements, SCHEPER-HUGHES (1997, p.219) argues that "there is no need for more collaborations between qualitatively-trained anthropologists or demographers in which the realm of the social is reduced to a set of reified and lifeless variables." She calls for a "praxis-oriented, critically applied and politically engaged anthropology to illuminate the complex and multifaceted ... dilemmas of vulnerable populations ..." (SCHEPER-HUGHES,1997, p.219). [36]

South African policy-makers are not averse to revising policy and making strategic decisions on the basis of approaches which are essentially qualitative, for example, opinions of community leaders, consultants or philosophical arguments. For arguments to be regarded as credible, however, it is usually expected that they be backed by "hard" data. In SCHEPER-HUGHES's view, advocates of an "engaged" qualitative methodology, tend not to be mere servants of the policy-making processes. They are "free agents" in pursuit of a deeper understanding of the complex issues influencing women's lives and their reproductive histories. We live in an era where researchers often have neither the time nor the patience to delve into deeper matters, and where quantitative data is being reclaimed for the grim tales that they could tell about the conditions of women's lives and the struggles they encounter (OAKLEY, 2000). These statistics will be highly contested, however, as the questions researchers ask and the claims they make are interrogated more rigorously. [37]

The purpose of this paper has been to review some of the contemporary literature on the problems posed by surveys and the possibilities raised by different kinds of qualitative approaches for doing fertility research. My concern here has been to draw attention to the specific difficulties of the South African, and in particular Winterveld terrain for measuring and eliciting detail on social practices and experiences. The case studies were intended to show the different ways in which qualitative work can be pursued: (1) as linked to survey research in a micro-demographic study, (2) as investigative work arising from prior survey work or (3) as a critical and engaged attempt to explore familial and reproductive issues over an extended period of time. Although I am of the view that a fully-fledged qualitative approach has a lot to offer to contemporary South African demography, I do not anticipate that many demographers would actively seek to become qualitative specialists. South African demography has not been sufficiently unsettled by debate on choice of methods. It remains a discipline with staunch adherence to survey methods and macro-level statistical analyses. It is possible, though, that the frustration of policy-makers will increase as the meaningfulness of quantitative data and their usefulness in informing social interventions become more frequently questioned. It will be in a context of questioning and rethinking paradigms and politics that the debate on "micro-macro mixes" and "engaged" qualitative methodologies will become more relevant. [38]

1) The Bantustans were, in apartheid South Africa, the reserves or homelands that the state created to accommodate the majority of the black population. They were established in 1951 in terms of broad policy to racially segregate South Africa. The black population was further sub-divided into 10 ethnic groups, with each group being allotted its own Bantustan. <back>

2) Despite changed politics, reference to apartheid created racial classifications is still common practice in contemporary South African social science research. In the previous era, South Africans were registered from birth as members of one of the four following racial groups: whites, coloureds (i.e. those of "mixed" largely white-black heritage), Indians and Africans (the black majority indigenous population). <back>

3) The micro-survey showed the backgrounds of the women to be quite diverse: 65% of women were migrants born elsewhere, having lived in the area for varying periods of time, with 35% having been born there. About 88% of those interviewed were tenants and 12% were plot-owners. The predominant languages spoken (of this sample) were Northern Sotho (18.4%), Zulu (18.1%), Shangaan (11.9%), Setswana (10.9%), Ndebele (10.6%), Southern Sotho (8.5%), Swazi (6.8%), Xhosa (6.5%) and other (8.3%). <back>

4) BLEDSOE et al. do not give their approach a particular name. I use this name merely because it appears to my mind to best summarise what it entailed. <back>

Anderson, Barbara A. (2003). Fertility, poverty and gender. In Department of Social Development (Ed.), Fertility. Current South African Issues of Poverty, HIV/AIDS and Youth. Seminar Proceedings (pp.27-56). Pretoria: HSRC.

Axinn, William G.; Fricke, Tom E. & Thornton, Arland (1991). The micro-demographic community-study approach. Improving survey data by integrating the ethnographic method. Sociological Methods and Social Research, 20(2), 187-217.

Blanc, Ann K.; Wolff, Brent; Gage, Anastasia J.; Ezeh, Alex C.; Neema, Stella & Ssekamatte-Ssebuliba, John (1996). Negotiating reproductive outcomes in Uganda. Calverton, MD: Macro International & Institute of Statistics and Applied Economics, Kampala, Makerere University.

Bledsoe, Caroline; Banja, Fatoumatta & Hill, Allan G. (1998). Reproductive mishaps and Western contraception: An African challenge to fertility theory. Population and Development Review, 24(1), 15-57.

Bongaarts, John (1978). A framework for analyzing the proximate determinants of fertility. Population and Development Review, 4(1), 105-132.

Brown, Barbara (1987). Facing the "black peril": The politics of population control in South Africa. Journal of Southern African Studies, 13(2), 256-273.

Caldwell, John C. (1985). Strengths and limitations of the survey approach for measuring and understanding fertility change: alternative possibilities. In John G. Cleland & John Hobcraft (Eds.), Reproductive change in developing countries. Insights from the World Fertility Survey (pp.45-63). Oxford: Oxford University Press.

Caldwell, John C. & Caldwell, Pat (1993). The South African fertility decline. Population and Development Review, 19(2), 225-261.

Department of Health (2004). South African Demographic and Health Survey 2003, Preliminary Report. Pretoria: Department of Health.

Du Plessis, Gretchen (1995). The development of demography in South Africa: A relevant social science or a tool for counting subgroups? Southern African Journal of Demography, 5(1), 52-59.

Du Plessis, Gretchen (1996). Reproductive choice and motivation in South Africa, 1987-1989. Southern African Journal of Demography, 6(1), 33-47.

Fricke, Tom (1997). The uses of culture in demographic research: a continuing place for community studies. Population and Development Review, 23(4), 825-832.

Greenhalgh, Susan (1995). Anthropology theorizes reproduction: integrating practice, political economic, and feminist perspectives. In Susan Greenhalgh (Ed.), Situating fertility: Anthropology and demographic inquiry (pp.3-28). New York, NY, USA: Cambridge University Press.

Greenhalgh, Susan (1997). Methods and meanings: Reflections on disciplinary difference. Population and Development Review, 23(4), 819-824.

Harriss, John & Moore, Mick (1984). Development and the rural-urban divide. London: Frank Cass.

Horton, Hayward D. (1999). Critical demography: The paradigm of the future? Sociological Forum, 14(3), 363-367.

James, Deborah & Kaufman, Carol (1997). The reproductive consequences of shifting ethnic identity in South Africa. Paper presented at the International Union for the Scientific Study of Population seminar on Cultural Perspectives on Reproductive Health, Johannesburg, South Africa, 16-19 June 1997.

Kabeer, Naila (1994). Reversed realities. Gender hierarchies in development thought. London: Verso.

Kaufman, Carol (1998). Contraceptive use in South Africa under apartheid. Demography, 35(4), 421-434.

Kertzer, David I. (1997). Quantitative and qualitative qpproaches to historical demography. Population and Development Review, 23(4), 839-846.

Kertzer, David I. & Fricke, Tom (Eds.) (1997). Anthropological demography. Toward a new synthesis. Chicago: University of Chicago Press.

Klugman, Barbara (1996). Demographics and population policy. Chapter 1. In David Harrison (Ed.), The South African health review (pp.3-10). Durban: Health Systems Trust.

Knodel, John (1997). A case for nonanthropological qualitative methods for demographers. Population and Development Review, 23(4), 847-853.

Marshall, Gordon (1994). Oxford concise dictionary of Sociology. Oxford: Oxford University Press.

Mostert, Willem P. (1990). Recent trends in fertility in South Africa. In Willem P. Mostert & Johann M. Lotter (Eds.), South Africa's demographic future (pp.63-73). Pretoria: HSRC.

Moultrie, Tom A. & Timaeus, Ian M. (2001). Fertility and living arrangements in South Africa. Journal of Southern African Studies, 27(2), 207-223.

Murray, Colin (1987). Displaced urbanisation: South Africa's rural slums. African Affairs, 86(344), 311-329.

Naidoo, Kammila (2002). Reproductive dynamics in the context of economic insecurity and domestic violence: a South African case study. Journal of Asian and African Studies, 37(3-5), 376-400.

Ndegwa, David G. (1996). International thinking on population policies and programmes from Rome to Cairo: Has South Africa kept pace? Southern African Journal of Demography, 6(1), 49-56.

Nkau, Poppy (1999). Fertility trends in South Africa. Unpublished report. Pretoria: HSRC.

Ntsaluba, Ayanda (1998). South Africa demographic and health survey. Preliminary report. Pretoria: Department of Health.

Oakley, Ann (2000). Experiments in knowing: Gender and method in the social sciences. New York: The New Press.

Obermeyer, Carla A.(1997). Qualitative methods: A key to a better understanding of demographic behaviour? Population and Development Review, 23(4), 813-818.

Rao, Vijayendra (1997). Can economics mediate the relationship between anthropology and demography? Population and Development Review, 23(4), 833-838.

Scheper-Hughes, Nancy (1997). Demography without numbers. In David I. Kertzer & Tom Fricke (Eds.), Anthropological demography. Toward a new synthesis (pp.201-222). Chicago: University of Chicago Press.

Simone, AbdouMaliq (1998). Winterveld. In Department of Constitutional Development (Ed.), Case studies on Local economic development and poverty (pp.272-299). Pretoria: Isandla Institute.

Swartz, Leon (2003). Fertility transition in South Africa and its impact on the four major racial groups. In Department of Social Development (Ed.), Fertility. Current South African issues of poverty, HIV/AIDS and youth. Seminar Proceedings (pp.7-26). Pretoria: HSRC.

United Nations (1995). Women's education and fertility behaviour. Recent evidence from the Demographic and Health Surveys. Population Division. ST/ESA/SER.R/137.

Xiphu, Twiggs (1998). NHISSA. Newsletter of the NHISSA Committee. Pretoria: Department of Health.

Kammila NAIDOO lectures in the Department of Sociology, University of Pretoria, South Africa. Her current research interests are in the areas of poverty, fertility and gender relations. This paper draws partly on her PhD fieldwork and research in Sociology completed at the University of Manchester, UK.

Department of Sociology
University of Pretoria
Lynnwood Road
Pretoria
0002
South Africa

Naidoo, Kammila (2007). Researching Reproduction: Reflections on Qualitative Methodology in a Transforming Society [38 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 9(1), Art. 12, http://nbn-resolving.de/urn:nbn:de:0114-fqs0801121.

Forum Qualitative Sozialforschung / Forum: Qualitative Social Research (FQS)

ISSN 1438-5627

Creative Commons Attribution 4.0 International License