Page 153 - Contributed Paper Session (CPS) - Volume 6

CPS1852 Leonard KA

Birth order and birth weight in Uganda: a
multilevel analysis of DHS data
Leonard K. Atuhaire
School of Statistics and Planning Makerere University Kampala, Uganda

Abstract
The ‘well-known’ influence of birth order on birth weight has, to our
knowledge, not been established in Uganda. In fact a recent study concluded
that birth order had no effect on (low) birth weight. DHS data, with thousands
of records that include both the birth order and the birth weight, with
matching mother records that include many maternal variables that are known
to influence birth weight, provide invaluable material to investigate the
influence of birth order on birth weight while adjusting for the effect of
maternal factors. However, the nature of DHS data is such that each woman
record is associated with several child records. Any analysis that includes both
mother and child factors has to take into account this hierarchical structure of
the data. A natural approach is to apply multi-level models, with the child as
the first level and the mother as the second level. This paper examines the
relationship between birth order and birth weight using the 2016 Uganda DHS.
A multilevel linear model for birth weight in kilograms and a multilevel binary
logistic model for the binary outcome (low birth weight) are fitted. The results
show that after adjusting for mother characteristics and other child
characteristics, increasing birth order is associated with increasing birth weight
as well as reduced incidence of low birth weight. The results also show
justification for multilevel modelling.

Keywords
Multilevel linear regression; Multilevel logistic regression; Random effects

1. Introduction
The ‘well-known’ influence of birth order on birth weight (Seidman et al.,
1988; Diamond et al., 2001; Côté et al., 2003) has to our knowledge not been
established in Uganda. In fact a recent study (Bayo et al., 2016) concluded that
birth order had no effect on (low) birth weight. DHS data, with thousands of
records that include both the birth order and the birth weight, with matching
mother records that include many maternal variables that are known to
influence birth weight, provide invaluable material to investigate the influence
of birth order on birth weight while adjusting for the effect of maternal factors.
However, the nature of DHS data is such that each woman record is
associated with several child records. Any analysis that includes both mother

142 | I S I W S C 2 0 1 9