Page 89 - IJPS-10-4

International Journal of
Population Studies Migration and child mortality estimation

Table 4. Paired sample t‑test results between NME and CME higher in rural regions compared to urban regions before
for urban and rural regions the 2010s (Kimani-Murage et al., 2014; Kenya National
Bureau of Statistics (KNBS), & International Classification
Child mortality rate Mean Std. error p‑value
(NME minus CME) Function (ICF), Macro, 2015). Previous studies from other
Urban region countries also indicate a significant rural-urban gap in
child mortality, with urban areas enjoying some advantages
q −4.292 1.009 0.002
1 0 over rural regions (Bocquier et al., 2011; Brockerhoff, 1994;
q −3.63 0.864 0.002
4 1 Issaka et al., 2017; Yaya et al., 2019). Factors such as access to
q −7.600 1.787 0.002
5 0 and utilization of health services, including immunization,
Rural region the use of oral rehydration therapy for diarrhea treatment,
q −0.336 0.201 0.106 and the provision of healthcare services by professionals
1 0
q −0.292 0.175 0.111 in urban areas, are cited as key contributors to the rural-
4 1
q −0.597 0.356 0.106 urban child mortality gaps (Govindasamy et al., 1993).
5 0
Notes: (i) q , q , and q refer to infant mortality, one-to-four mortality Ideally, if the rural region were experiencing a higher
1 1 4 1
5 0
rate, and under-five mortality rate, respectively. (ii) NME refers to the mortality rate than the urban region, we would expect
mortality estimates obtained from non-migrant women, and CME an underestimation in the rural region due to urban-to-
refers to the mortality estimates obtained from non-migrant women
and in-migrant women combined. (iii) The mean for each indicator rural migrant women (Schmertmann & Sawyer, 1996). On
was estimated from 36 rates (six surveys with six age groups each [age the contrary, our study demonstrates that the migration
group 15 – 19 was dropped due to the very small number of women of women from urban to rural regions led to a slight
who had a child in this group]). overestimation of infant, child, and under-five mortality
Abbreviations: NME: Non-migrant mortality estimate; rates in the rural region. This slight overestimation of
CME: Combined mortality estimate.
child mortality rates in rural regions due to deceased
children born to migrant women from urban regions
presence of two sets of mortality data: One for children who suggests the possibility of return migrants with higher
died in the migrated women’s previous place of residence
and another for children who died in their current place of mortality experiences. Previous studies have shown that
residence. In the survey data from Kenya, the date of death many migrants from rural to urban regions often end
of a child is usually provided in months and years, while the up living in deplorable conditions, such as slums with
date of migration is given in years. This discrepancy makes inadequate social services (Madise et al., 2003; Van De
it challenging to determine the geographical region where Poel et al., 2007). After failing to realize their economic
some children of migrant women died. Furthermore, the dreams in urban regions, these individuals return to their
Brass indirect technique typically requires a large number original rural regions. Our results concur with the study
of births and deaths because estimates from small samples by Andersson & Drefahl (2017), which reveals that return
are highly susceptible to random errors. Consequently, we migrants usually have elevated mortality. Moreover, our
were not able to compute mortality estimates for children study aligns with the study by Otieno Onyango et al. (2011),
born to migrant women alone due to the limited number which shows that children born to migrant mothers have
of deceased children born to this group. To address these a higher risk of mortality than those born to non-migrant
challenges, our analysis compared mortality estimates for women. Supporting evidence also comes from Issaka et al.
deceased children born to in-migrant and non-migrant (2017), who, by pooling data from 27 sub-Saharan African
women combined with those born to non-migrant women. countries, showed elevated under-five mortality for rural
The difference between the two estimates was attributed to non-migrant (by 40%), rural-to-urban migrants (by 43%),
migration. and urban-to-rural migrants (by 20%) in comparison to
urban non-migrants.
The results of the pairwise t-test demonstrate that
migration led to an overestimation of the infant, child, and We also noted some anomalies in the results of the
under-five mortality rates in both rural and urban regions. 1993 survey. These results indicate an underestimation of
This impact was significant in the urban region but not in child mortality rates in urban regions. This discrepancy
the rural region. Overestimation of mortality rates in the can be attributed to the structural adjustment program
urban region suggests that rural-to-urban migrant women implemented in Kenya during the early 1990s, which
had experienced relatively higher mortality in the rural prompted a significant urban-to-rural migration due to
region before migrating compared to urban non-migrant economic hardships in urban areas (Government of Kenya,
women. Evidence to support this explanation comes from 1996). Our findings have several implications for both
earlier studies in Kenya that suggest child mortality was programming and monitoring. First, there is an impact on

Volume 10 Issue 4 (2024) 83 https://doi.org/10.36922/ijps.1837