Page 56 - IJPS-3-2

Zewdie S A and Adjiwanou V

only associated with 3.5% higher odds of infant death than those municipalities where poverty was low. Similarly, infants
living in municipalities where the average years of education of women was higher had a better likelihood of survival
irrespective of the education level of their own mother. They were about 16% less likely to die than infants living in areas
where there was less education of mothers.
Considering the effect of HIV on mortality of infants, it is apparent that its coefficient is positive and significant. Infants
in provinces with high HIV prevalence were 12% more likely to die than infants in other provinces after controlling other
factors in the model.
The sex effect is also positive and significant at p<0.001 level. More specifically, boys were 8% more likely to die
during their first year of life than girls, all else being equal. Similarly, age of the child significantly affected the mortality
of the child in that children who survived the first month after birth were less likely to die than those who were less than
one month old by about 7%.
In terms of mother characteristics, the results in Table 3 show that mothers’ age at the birth of the child was not
statistically significant. All the other independent variables associated with mothers were significant, which include years
of education, employment status, birth order, marital status and population group.
In accordance with the literature, the more education a mother gets, the less likely the infant dies. Infant whose mothers
completed secondary and higher education were 7% and 17% respectively less likely to die than those children whose
mothers had no education or only had primary education. In addition, infants of higher birth orders, infants born to mother
who were never married or single mothers, and infant born to mothers from African population group had a greater chance
of dying than their counterparts of lower birth order, born to mothers who were not single and born to mothers from non-
African population group. Specifically, an infant of the fourth or higher birth orders was associated with 15% higher odds
of dying than first born infants, whereas an infant of the third birth order was associated with 8% of higher odds of dying.
No difference was found in infants mortality between the second birth order and the first birth order. Similarly, children
born to Black women or born to not-married women were associated with 13% and 5% higher odds of dying before age 1
in comparison with children born to non-Black or born to married women.
It is also worthy to see the distribution of the predicted log-odds of infant deaths in the country to see how the multilevel
model performs in estimating infant mortality. The predicted log-odds of death of the municipalities and provinces are
shown in Figure 1 and 2 below. It can be observed that municipalities such as City of Cape Town, Liansburg, Mossel Bay
and Nama Khoi had the lowest level of infant odds of death, while Disobotla, Mamusa, Msukaligwa and Ingwe scored

-1.8 Msukaligwa, MP Ditsobotla, NW
Mamusa, NW
Ingwe, KZN

Predicted log-odds of death -2 -2.2

Nama Khoi, NC
Mossel Bay, WC
-2.4 Laingsburg, WC
City of Cape Town, WC
0 50 100 150 200 250
Municipality

Figure 1. Predicated log-odds of infant death across the municipalities of South Africa
International Journal of Population Studies | 2017, Volume 3, Issue 2 49