The impact of the number of siblings on child health

Project description

Economic, demographic and sociological as well as epidemiological literature has raised concerns regarding the well-being and life chances of children raised in large families. According to the resource dilution model, in larger families the parental financial support as well as personal attention are distributed across a larger number of siblings, and therefore the parental investments per each child may be lower. Hence, parents face trade-offs between the quantity and “quality” of children when making decisions regarding the size of their family, with the key “quality” dimensions defined as education attainment and health. Stronger effects of family size can be expected in poorer families or families of lone parents whose resources of time and income are more restricted.
Previous empirical research has focused mainly on one of the two dimensions of the impact of family size on child “quality”, i.e. on educational chances of children. The majority of these studies seem to confirm the predictions of a resource dilution model. Much less attention has been paid so far to the effects of growing up in a large family on child health. This is an important gap in the literature because if parents with more children devote less attention towards monitoring child activities and assuring that their children adopt a healthy life style, the family size may have a long-lasting effect on the quality and duration of children’s lives.
The aim of this project is to examine the impact of the number of siblings on child physical and mental health. Specifically, it will assess whether children raised in larger families differ from those in smaller families in terms of: consumption of prescribed drugs, use of hospital care and mortality, as compared to children in smaller families. The magnitude of effects of family size will be compared across various family types and different income strata. These research questions will be addressed by using high quality data from Swedish Registers that cover the whole population and allow detailed analysis.
Previous research has been criticised on methodological grounds as it has applied methods that reveal associations rather than causal effects. Testing the hypothesis on trade-off between family size and child “quality” is challenging because decisions to have another child may be driven by the same factors which simultaneously have impact on the offspring’s education or health and are difficult to observe and measure directly in the data. For example, parents with better cognitive skills may prefer to have smaller families and simultaneously invest more in education or health of each child. Hence, even in the absence of true causality, there may be a strong negative correlation between family size and child “quality”. This means that the results from standard regression models might lead to misleading conclusions. Indeed, the most recent studies using econometric methods which deal with endogeneity of the family size reveal that the seemingly well documented correlation between the family size and child education is spurious.
In this project, the impact of the number of siblings on child health will be examined with a research design which handles the problem of non-randomness of family size. We will consider the number of siblings in the family as the treatment variable, and the measure of health of older siblings of this child as the outcome variable. The standard approach to infer the effects of causes is to conduct controlled randomised experiments, but in the social sciences, such experimental designs are frequently infeasible for technical and/or ethical reasons. In lieu of opportunities for carrying out randomized experiments, social scientists have started to utilize quasi experiments for causal inference. In this strand of literature, an instrumental variable model is a commonly chosen approach. Another increasingly popular design in observational studies is propensity score matching, which has only recently been adopted in research on the consequences of family size choices. Each of these two approaches has its advantages and limitations. In this project, the robustness of our results will be checked by applying both an instrumental and a propensity score matching design.
This project will address the aforementioned challenges by taking advantage of a unique database made available at the Umeå SIMSAM Data Lab which combines Swedish registers that cover the whole Swedish population over a number of decades. These data capture all the multiple births that took place since 1960. They provide detailed information on health of both parents and their children. In the Prescribed Drug Registry, one can identify prescribed and discharged medicines, and in the National Patient Registry, hospitalization data are available to us. Based on the data available in these registers it is possible to measure the incidence of receiving the prescriptions and the number of dispensed defined daily doses, as well as incidence rates of admissions to hospital and the length of stay in hospitals.