What is wellbeing for rural South African women? Textual analysis of focus group discussion transcripts and implications for programme design and evaluation

The study interrogates whether commonly used unidimensional measures of wellbeing (e.g., life satisfaction or happiness) are appropriate for assessing population welfare in low- and middle-income settings, particularly among poor rural South African women. Drawing on the mental wellbeing (MWB) framework that distinguishes hedonia (affect, life and domain satisfaction) from eudaimonia (psychological and social wellbeing), the paper argues that interventions may differentially affect these facets, potentially in opposite directions, and that unidimensional indices may obscure such patterns. The purpose is to explore how women in peri-urban Limpopo construct and describe wellbeing, and to derive implications for programme design, targeting and evaluation aligned with Sustainable Development Goal 3. The study is important because most existing evidence and measures are developed in high-income contexts, with limited attention to gendered and cultural variations in self-construal that shape wellbeing perceptions and policy-relevant mechanisms.

The paper situates its inquiry within a multidisciplinary literature on wellbeing. It adopts the mental wellbeing taxonomy (Joshanloo et al., 2016), distinguishing hedonia (positive/negative affect, life and domain satisfaction) and eudaimonia (psychological wellbeing: autonomy, environmental mastery, positive relations; and social wellbeing: contribution, integration, actualisation, coherence, acceptance). While hedonia and eudaimonia can be correlated, accumulating evidence supports treating them as distinct constructs. Self-construal (individuated, relational, collectivist) shapes wellbeing concepts and responses. African contexts, influenced by Ubuntu, often reflect relational interdependence distinct from East Asian collectivism and Western individualism. Prior large-N studies in South Africa using unidimensional measures report associations of life satisfaction with income, neighbours’ income, marital status, pensions, education, health, and crime, but provide limited insight into mechanisms or eudaimonic dimensions. Emerging qualitative work in South Africa (e.g., adolescents, older adults) suggests multidimensional constructs encompassing subjective, psychological, and social wellbeing. The paper highlights gaps: few grounded explorations of adult women’s MWB in African contexts and limited attention to how relational self-construal modifies domain–wellbeing linkages.

Design: Exploratory qualitative study using focus group discussions (FGDs) and quantitative textual analysis (cluster and correspondence analysis of word co-occurrences) to elicit and structure socially shared perceptions of mental wellbeing. Setting and participants: Conducted in peri-urban villages of Greater Tubatse municipality, Limpopo Province, South Africa, among the baPedi population targeted by the Intervention with Microfinance for AIDS and Gender Equity (IMAGE). Thirteen FGDs (2006–2008) with 79 Pedi women aged 22–65 from the three lowest income quintiles; basic literacy common. Sampling was purposive, recruiting IMAGE clients (varying exposure: none to 5 years) and non-clients from the same socio-economic background via natural groups (neighbour or loan co-recipient groups). Sessions held indoors for privacy; each about three hours with ~6 participants; simultaneous translation used. Data collection: Four components per FGD: introduction; brainstorming; two narrative activities; and conclusion. - Brainstorming prompt: “When I say ‘happiness’ what first comes to your mind?” Translated into sePedi (“Ke thabile ga...”). Participants reflected silently, then shared associations; facilitator emphasised no right/wrong answers. - Life histories graph: Participants plotted mental wellbeing across life stages (childhood, adolescence, marriage, motherhood/parenting, old age/grandmotherhood, death) on a 3–4 point smiley-face scale, narrating key events. - The day before graph: Adapted Day Reconstruction Method to FGD; participants recounted prior day’s activities across waking hours (06:00–21:00), plotting associated emotions on the same scale. Probing explored mechanisms linking experiences to happiness/wellbeing. Positionality: The author, a white, highly educated outsider, spent extended periods in the field over four years, learned basic sePedi, and worked closely with local research assistants to mitigate positionality effects and build rapport within relational dyadic networks. Data preparation and analysis: FGDs were translated and transcribed into ~52,000 words of English text. Analysis used Alceste software to perform: - Descending hierarchical classification (cluster analysis) on lexeme–sentence co-occurrence matrices, iteratively partitioning into homogeneous themes. Chi-squared statistics indicate strength of word–theme association (no fixed significance thresholds). - Correspondence analysis to map themes and participant groups in a two-dimensional vector space; distances reflect association strengths; axes interpreted as latent structural dimensions of discourse. - Distributional analysis of wellbeing-related words (adjectives/nouns indicating moods or cognitive assessments) across themes. Ethics: Approvals obtained from University of the Witwatersrand, London School of Hygiene and Tropical Medicine, and London School of Economics; informed consent obtained; village leadership permissions secured.

- Five themes of discourse emerged: 1) Female socialisation: Life-course transitions and socially codified interactions (e.g., marriage), highlighting low autonomy in early marital experiences and reliance on peers to navigate sexuality; some accounts of intimate partner violence and extreme unhappiness. 2) Parenting: Centrality of mother–child dyad in a context of crime and HIV; feelings of powerlessness when children engage in wrongdoing; legislation limiting corporal punishment contributed to perceived reduced competence; HIV sub-theme showed awareness, competence, and proactive guidance. 3) The community: Family, neighbours, peers, and church as key networks; selective trust; breaches of confidentiality damage wellbeing; conflict risks exclusion from support; helping peers produces positive affect and a sense of empowerment; networks have ambivalent effects (supportive vs. sources of jealousy/strained comparisons). 4) The household: Daily chores and physical demands (e.g., fetching water); physical health and the ability to accomplish tasks foster wellbeing; tiredness co-occurs with happiness via achievement; caregiving roles are relational; high burden and stress linked to childcare routines; socialising sub-theme captured affective connections. 5) Mulier oeconomica: Women as economic agents (employment, enterprise, loans); income enables provision for children’s needs and generates satisfaction; unequal spousal financial cooperation causes unhappiness; gratitude/spirituality linked to achieving economic goals. - Latent dimensions (correspondence analysis): • Y-axis (self-construal): From relational role-relationships (female socialisation; parenting; community) to more individuated roles (household decision-making; mulier oeconomica as entrepreneur/breadwinner). • X-axis (empowerment): From disempowerment (female socialisation; household) through negotiated power (community; parenting) to transformative agency (mulier oeconomica). - Relational self-construal predominated, distinct from Western individuated and East Asian collectivist models; women navigated dyadic networks and role obligations, yet shifted toward individuated agency in economic roles. - Wellbeing word distributions by theme: • Parenting: Highest extreme negative affect; 43% of MWB words were stressed, worried, or angry; rising to 70% when ‘unhappy’ included. • Household: High proportion of negative states; after parenting, the highest share of unhappy mentions (50%); more neutral states than female socialisation. • Community: ~60% positive MWB words; ~20% negative (angry/stressed), reflecting ambivalence of networks. • Mulier oeconomica: Highest proportion of positive MWB states (71%); ‘satisfied’ mentions statistically significant only in this theme. - Sample/context statistics: 79 Pedi women (22–65 years); IMAGE exposure distribution: none N=18 (23%), <1 year N=23 (29%), 2–3 years N=24 (30%), 4–5 years N=14 (18%). - Policy-relevant mechanisms: Autonomy and environmental mastery often enacted relationally (via peer identity and networks); networks are inescapable and can be both empowering and oppressive; income’s positive impact is channelled through social contribution and relational mastery (provider role).

The findings align with some large-N associations (e.g., parenting linked to more negative affect; networks and income-related activity linked to higher life satisfaction) but add mechanism-rich nuance by revealing a relational self-construal underpinning women’s mental wellbeing. Women’s wellbeing is anchored in fulfilling role-relationships (mother, wife, entrepreneur) within dyadic, reciprocal networks. This relationality modifies how domains affect different MWB dimensions: marriage and early socialisation often undermine autonomy, positive relations, and social integration (especially in customary unions), while community networks provide both acceptance/integration and risks (jealousy, breaches of trust), leading to offsetting effects that unidimensional life satisfaction measures may mask. Economic roles (mulier oeconomica) heighten agency and satisfaction, demonstrating that individuated and relational selves can coexist and shift with context. Physical work can decrease hedonic states while increasing eudaimonic feelings of accomplishment and purpose. Overall, multidimensional MWB metrics that separately track autonomy/agency, environmental mastery/competence, and meaningful relationships (including social contribution, integration, acceptance) yield clearer, actionable insights for policy and programme evaluation in relational contexts. Empowerment interventions should emphasise relational forms of agency (peer support, community mobilisation) and anticipate both positive and negative network effects, designing components that enhance social contribution and broader integration.

Unidimensional MWB measures focused on hedonia are insufficient to inform policy in contexts where wellbeing is constructed relationally. Textual co-occurrence analysis of FGDs identified three additional components—agency (autonomy), environmental mastery (competence), and meaningful relationships with others (positive relations and the social wellbeing facets of contribution, integration, and acceptance)—that should be included alongside happiness/life satisfaction in policy evaluation. Networks are not uniformly beneficial; because they are inescapable, they can both empower and oppress, producing opposing effects across MWB dimensions that may cancel out in single indices. Multidimensional measures sensitive to relationality will better detect discordant impacts, guiding targeted, contextually appropriate interventions. Future research should examine MWB constructs across diverse sub-Saharan groups and urban populations, and empirically map how specific empowerment components translate into MWB gains, leveraging validated instruments such as the MHC-SF or GHQ-12 where culturally appropriate.

- Elicitation approach prioritised simple, open-ended prompts to surface shared, socially produced meanings; deeper individual probing was limited by design. A small number of in-depth interviews (n=3) with other women suggested similar content and were discontinued. - Translation and interpretation may risk meaning loss; mitigated by providing verbatim simultaneous interpretation alongside translator’s version and discussing discrepancies. - Findings derive from FGDs in peri-urban Limpopo among poor Pedi women (natural groups), potentially limiting generalisability to other demographics (e.g., metropolitan populations) though consistency with related Southern Bantu evidence supports broader relevance. - Data were collected in 2006–2008; the author argues concepts of self and MWB are relatively stable, with recent studies showing consistency, but temporal changes cannot be entirely excluded. - Textual analysis relies on co-occurrence structures; while powerful for theme discovery and distancing researcher bias, it depends on quality of transcription/translation and does not quantify causal relationships.