Page 60 - IJPS-7-2

International Journal of
Population Studies Child mortality by residence in Ethiopia

3.3. Results from decomposition analyses urban differential of 39 deaths/1000 children. There would

The multilevel analyses have provided evidence for be 39 deaths/1000 children change in rural areas when
the existence of a statistically significant higher risk of applying the urban areas coefficient and characteristics to
child death in rural areas as compared to those living in rural areas’ behavior/characteristics. In the rural-urban
the urban areas in Ethiopia. However, the rural-urban decomposition analysis, child size at birth, birth order,
residential gap and the intra-rural and intra-urban type of sanitation facilities, and wealth status contributed
gaps are not known yet. For this purpose, the present to explaining the rural-urban gap in child mortality. On
study employed Blinder-Oaxaca decomposition analysis the other hand, child sex, religion, household size, and
based on the place of residence and urban/rural wealth regional categories contributed to widening the child
status, separately. Table 3 presents the decomposition of mortality gap.
the rural-urban inequalities in child mortality grouped Table 4 depicts the decomposition of the intra-rural
by place of residence. The rural-urban decomposition inequalities in child death grouped by rural wealth status.
analysis showed the mean predictions by groups and The intra-rural decomposition analysis revealed that the
their difference. For the sake of understanding, we mean proportion of child death was 76 and 64 deaths/1000
converted mean prediction into proportion per 1000. The children for rural poor and rural non-poor, respectively.
decomposition result showed that the mean proportion of As a result, there would be 12 deaths/1000 children change
the child death was 71 deaths/1000 children for rural and in rural poor when applying the rural non-poor coefficient
32 deaths/1000 children for urban areas, resulting in rural- and characteristics to rural poor characteristics. The intra-

Table 3. Blinder‑Oaxaca decomposition of rural‑urban in risk of child death, EDHS, 2016
Child mortality Coefficient Standard error z statistic P‑value 95% confident Interval
Overall
Rural 0.071 0.003 22.750 0.000 0.065 0.077
Urban 0.032 0.005 7.080 0.000 0.023 0.041
Difference 0.039 0.006 6.970 0.000 0.028 0.049
Explained 0.017 0.060 0.280 0.777 −0.100 0.134
Unexplained 0.022 0.060 0.360 0.716 −0.095 0.139
Explained
Child sex −0.000 0.000 −0.750 0.451 −0.000 0.000
Child size 0.008 0.007 1.110 0.267 −0.006 0.022
Birth order 0.047 0.034 1.380 0.167 −0.020 0.114
Religion −0.009 0.009 −1.010 0.314 −0.027 0.009
Household size −0.032 0.022 −1.460 0.144 −0.074 0.011
Toilet type 0.027 0.038 0.700 0.481 −0.048 0.101
Wealth status 0.018 0.051 0.350 0.725 −0.081 0.117
Regional cat. −0.042 0.029 −1.430 0.154 −0.099 0.016
Unexplained
Child sex 0.000 0.005 0.100 0.923 −0.010 0.011
Birth weight 0.019 0.041 0.460 0.644 −0.061 0.098
Birth order 0.027 0.060 0.460 0.646 −0.090 0.145
Religion −0.020 0.044 −0.460 0.646 −0.106 0.066
Household size −0.033 0.072 −0.450 0.652 0.174 0.109
Toilet type 0.001 0.015 0.090 0.927 −0.027 0.030
Wealth status −0.015 0.023 −0.680 0.495 −0.060 0.029
Regional cat 0.019 0.051 0.370 0.709 −0.081 0.119
_cons 0.023 0.047 0.490 0.627 −0.070 0.115
Note: The models were based on 8,162 observations using logistic regressions, with 6,680 observations from rural samples and 1,482 observations from
urban samples.

Volume 7 Issue 2 (2021) 54 https://doi.org/10.36922/ijps.v7i2.392