Page 37 - IJPS-2-2

Mukesh Ranjan, Laxmi Kant Dwivedi, Rahul Mishra, et al.

and non-tribes. In the Cox model, the dependent variable is the length of exposure (from the date of
the initial observation to the date of the event). In our analysis the event variable is infant death,
which is coded 0 as “no infant death” and 1 as “infant death” (i.e., death occurred within 12 months
of birth). We have seen the impact of various covariates separately for tribes and non-tribes. We
tested the proportionality assumption by using Schoenfeld residuals and found that no variable vi-
olated the proportionality assumption. All analyses were performed using Stata13.0.

3. Results

3.1 Sample Description
Table 1 shows that around 54,584 live births occurred to 119,534 women from January 2004 (i.e.,
the reference period for tracking births in the study) to the date of the DLHS-III in 2007–2008. In the
DLHS-III in 2007–2008, 86% of the population in Central and Eastern India resided in rural areas
and most of them (around 90%) were Hindus. For tribes, 95% of their populations were based in
rural areas, while the corresponding figure for non-tribes was 80%. This indicates that there was a
large homogeneity in culture among tribes due to sharing of similar types of place of residence. The
wealth differentials between tribal and non-tribal populations showed non-tribes as better off than
tribes, with 88% of tribe members found to be poor but only 53% for non-tribes; the percentage of
the rich among non-tribes was seven times more than that among tribes. This shows that a wide dis-
parity existed in terms of the distribution of wealth between tribes and non-tribes.
In our sample, around three-fourths of tribe women were illiterate, while for non-tribes it was less
than 50%. The proportion of tribe women with an educational attainment of high school or above
was less than 2%, smaller than for non-tribe women, which was more than 8%. We included one
program factor (i.e., place of pregnancy registration) in our analysis. In India (including the Central
and Eastern Indian region), private health care centers have much better facilities than public
ones but with higher cost of accessibility. In our sample, we found that the proportion of pregnancies
registered at a private healthcare center was higher among non-tribes than among tribes.
3.2 Cumulative Hazards Rates

Figures 1 and 2 represent the Nelson-Aalen cumulative hazards rate curve for tribal and non-tribal
populations for the Central and Eastern regions of India and for each state. The curves reveal that
infants in tribes were at higher risk of death in the first year of life both in the entire region and
within each state. Variation across states shows that in Madhya Pradesh the cumulative hazards rates
of both groups were close, meaning that the risk of infant death is similar between tribes and
non-tribes. In the remaining three states, the hazards rates were higher for tribes and remained higher
during the first year of life. This indicates that the disparity in risk of infant death existed during the
first year of life between the tribal and non-tribal populations in the states of Jharkhand, Odisha, and
Chhattisgarh. The largest disparity in hazards between tribes and non-tribes was found in Chhattis-
garh (Figure 2).
Figure 3 highlights the hazards rates by age of the child after birth for both tribes and non-tribes.
The pattern of curves was quite similar for both groups. The curve for both groups (tribes and
non-tribes) shows that in the very early days of birth, the risk of death was quite high. Later, the
mortality risk decreased with increased age of the child till 4 months, the level of risk remained con-
stant between 5 to 10 months for both groups, and increased thereafter. However, the hazards curve
for tribes remained above the hazards curve of non-tribes for the entire period.
The log rank test confirmed that children of tribes in the region were at higher risk of death than
non-tribes in their first year of life as there was statistically significant difference between the sur-
vival distributions of tribal and non-tribal populations (Table 2).

International Journal of Population Studies | 2016, Volume 2, Issue 2 31