Exploring inequalities in child cognitive ability, psychological well-being and risky health behaviours

Abstract

The aim of this thesis was to analyse household inequalities in child and adolescent outcomes. Using the National Child Development Study, British Cohort Study and Millennium Cohort Study, the first empirical chapter estimated the extent of socioeconomic inequality in child cognitive ability, investigated if the magnitude of these inequalities had changed significantly over time, and decomposed the inequality into its contributing factors. Results showed substantial socioeconomic inequalities in child cognitive ability. There was limited evidence that the magnitude of the relationship had changed over time. Income and parental occupational classification accounted for the majority of income related socioeconomic inequality, with smaller roles for maternal education and family size. The second empirical chapter estimated the impact of both family size and birth order on child cognitive ability and psychological well-being, using the Millennium Cohort Study. Ordinary Least Squares models indicated a negative conditional association between family size and psychological well-being, but not cognitive ability. Two Stage Least Squares models, using two separate identification strategies, showed no causal effect of family size. For birth order, both Ordinary Least Squares and Nearest Neighbour Matching models showed substantial later born advantages for the certain subscales of psychological well-being, with this relationship in general not shown for cognitive ability. The third empirical chapter estimated the impact of both maternal labour market supply and non-standard work schedules on adolescent risky health behaviour, using the UK Household Longitudinal Survey dataset. Using a variety of panel data models, there was evidence of a small conditional association between maternal working hours and adolescent drinking, with this relationship not shown for smoking. Two instrumental variable strategies implemented to identify a causal effect were shown to be inappropriate for the research question. For the incidence of non-standard work schedules, there was little evidence of a conditional association for either risky health behaviour.