Menu
Menu
Menu
Menu
We used data from the Birth to Twenty Cohort study to understand children’s receipt of financial support from their fathers in a low income, Black community in urban South Africa. Specifically, we (1) described fathers’ financial support over the life course of children; (2) estimated survival probabilities of receiving support for all children and notreceiving support for children who experienced a parental union dissolution; and (3) identified factors that explained variation in the receipt of support after a union dissolution. Results suggest that most children received full or partial support throughout the life course. Furthermore, a high proportion of children received support after a union dissolution with much of the variation driven by pre-dissolution support, father’s education and the presence of extended kin.
Bt20 has been the longest running birth cohort study in Africa situated in the greater Johannesburg-Soweto municipality in South Africa (Sabet et al. 2009; Yach et al. 1991). The majority of families, most of whom were Black, came from socioeconomically disadvantaged circumstances. Bt20 was initiated as an observational, systematic study of human development, health and well-being, from birth extended through to young adulthood. Data collection covered a broad range of topics including anthropometric measures, nutrition, family composition, socioeconomic circumstances, childcare, parenting, cognitive development, and social experiences at home, school, and in the community. Prospective data collection began in the antenatal period and continued with approximately 21 follow up visits until age 23. Children born between April and June 1990 and resident for at least 6 months in the Soweto-Johannesburg municipality were enrolled into the study (n = 3273). The cohort included Black, White, Indian, and Colored children but we limited this analysis to only the Black children (n = 2568) who comprised the largest proportion of the cohort in line with the population distribution of the area. Even though data have been collected through age 23, this analysis used age 18 as the end point. While these data are not nationally representative, they offer some of the richest data on father involvement in the South African context, and therefore, are highly suitable for this analysis. Data in Bt20 on father involvement have been collected in two ways. Prospective data collected as part of household rosters to determine father co-residence, father contact and provision of financial support by fathers for most years of data collection. In addition, a retrospective questionnaire specifically focusing on father involvement across the child’s life course was administered at year 18. The questionnaires, most of which were answered by mothers, included detailed information on fathers’ co-residence with the child, extent of contact if not co-resident, provision of financial support, and other forms of interaction with the child for every year from birth until age 18. To both maximize our sample size and improve the validity of our measures, we used the retrospective data to supplement the prospective data but always privileged the prospective when it was available. There are two drawbacks that need to be acknowledged. One, most of the information about fathers came from mothers or other caregivers. Research from the US context has highlighted the potential biases in mothers’ reports, which consistently show underreporting of father involvement (Coley and Morris 2002). It is difficult to establish the extent of such bias in the Bt20 data but comparison of mothers’ reports of father contact over the life course and fathers’ reports of their own involvement (from the year 18 biological father questionnaires) suggested potential underreporting. Two, the use of retrospective data introduced problems associated with memory recall the farther back in time that data were sought. However, when we compared retrospective reports of father presence in the 0–2 year period with prospective data for the same time period, we found that 85 % of reports matched. Attrition over the course of the BT20 study has been about 30 %, mostly occurring during infancy and early childhood when women moved back to their rural homes after giving birth (Norris et al. 2007). A small number of children were lost to follow-up as a result of death. There have been very few withdrawals from the study. After removing non-Black children, the sample was 1,942 girls and boys followed up from birth to age 18, out of which, 1,557 were administered the retrospective father questionnaire. Table 1 shows descriptive characteristics of the analytical sample at the time of birth. Selected characteristics of analytical sample at time of birth (N = 1,557) A little more than a third of the cohort was composed of first births and the mean age of mothers at birth of the index child was 25.8 years. More than a third of all mothers were married or living together (a term used synonymously with cohabiting) with their partners. The majority of mothers had at least some secondary school education. We found a similar distribution for fathers on educational attainment though there was a sizeable missing proportion. The household wealth index used in this analysis was computed as quintile rankings of asset scores based on home ownership, access to regular electricity and ownership of car, TV, refrigerator, and phone. It ranged from 1 (very poor) to 5 (wealthy) and showed the highest proportions in the 2nd and 3rd quintile. Finally, the majority of households were classified as extended family structure although there were a sizeable number of records with missing data. Children’s receipt of financial support was treated as a dichotomous outcome (1/0) based on responses to the question, “In the past year, who was mainly responsible for the material support of the child?” To examine the timing of children receiving financial support from fathers, we used Kaplan–Meier estimation techniques to determine the survival probabilities of (1) receiving financial support from fathers for all children, and (2) not receiving financial support from fathers for those children who experienced a union dissolution. Although we recognized that the events of interest could recur, in these analyses we considered only the first observed event because of insufficient sample size. To examine correlates of father support provision post-union dissolution, we used a discrete time event history model. The child cohort was comprised of all children who had ever experienced a parental union dissolution before the age of 18. A child entered the cohort at the year of parental union dissolution. The dependent variable or event of interest occurred when a child received financial support from the father for the first time post-dissolution. Children who received support in the year of dissolution were included in the analysis and their odds of experiencing the event started at the year of dissolution. An observation was censored if the event did not occur by the age of 18 when the observation period ended or when the father died. Each child’s exposure time was divided into child-years starting at the time of parental union dissolution and consisting of 1 year intervals resulting in 3,777 child years of exposure. We used logistic models in SPSS to estimate the odds of children receiving financial support in the post-dissolution period. Paternal attributes included father’s age and educational level at time of birth of the child, which was also treated as a measure for employment potential. The maternal characteristics included mother’s age and education at time of birth, and whether she entered a new union following dissolution in the first 5 years after dissolution. Child characteristics included sex of child, and age at time of parental union dissolution categorized into four developmental stages (0–2, 3–5, 6–11 and 12–18) and entered as a continuous variable. Kin involvement was treated as a continuous variable measured by number of co-resident non-parental adults (data available in the household rosters). Control variables included household wealth at time of birth measured by quintile ranking of asset score based on ownership of household items (1–5) and whether father provided any support before or at the time of union dissolution. All covariates were time constant except number of co-resident non-parental adults which was treated as time varying and measured at the beginning of each period. In our quest for maximizing sample size by integrating retrospective data with the prospective data, the analytic sample was composed of only those children who “survived” in the study until year 18. It is, therefore, possible that those children who were lost to follow up might have had weaker links to their fathers which would, in turn, contribute to an overestimation of father involvement in our analysis. We examined this issue by comparing the means of duration of father contact for children who dropped out and those who did not by age of attrition. With the exception of two attrition periods: 6 months–2 years and 12–13, none of the differences were significant suggesting that our estimates of father contact in this analysis were not seriously affected by survivor bias.
N/A
