I will present work from the Big Data Theme component of the ‘A Better start’ National Science in New Zealand. We used linked records from birth registrations, the 2013 Census, and a national screen of 4-year-olds health and development (the B4 School Check) to determine the extent to which ethnic differences in BMI Z-scores and obesity rates could be explained by the differential distribution of demographic (e.g., age), familial (e.g parental education), area (e.g area deprivation), parental (e.g immigration status), and birth (e.g gestational age) characteristics across ethnic groups.
Overall, 15.2% of the sample were obese and mean BMI z-score was 0.66 (SD=1.04). The Oaxaca-Blinder decomposition demonstrated that the difference in obesity rates between Māori children and European children would halve if Māori children experienced the same familial and area level conditions as European children. If Pacific children had the same characteristics as European children differences in obesity rates would reduce by about one third, but differences in mean BMI Z-scores would only reduce by 16.1%.
The differential distribution of familial, parental, area, and birth characteristics across ethnic groups explain a substantial percentage of the ethnic differences in obesity, especially for Māori compared to European/Other children. However, marked disparities remain.