Page 31 - IJPS-3-2

Upadhyay AK et al.

observed between children from non-migrated households irrespective of whether had experienced an economic shock or
not. Therefore, the significant effect in our study can be safely considered as an actual effect of forced migration on child
cognition. Findings of the present study also indicate that the effects of forced migration on child cognitive well-being
were not mitigated by social support. One of the possible reasons for the lack of association between social support and
childhood cognitive well-being may be reverse causality, whereby households who were in a more adverse condition were
more likely to receive support from other individuals within their community or from relatives and friends. Kawachi and
Berkman (2001) reported that the protective effect of social support may not be uniform across society. To the best of our
knowledge, this is the first study that examined the causal association between forced migration and child cognitive well-
being in India.
When interpreting our findings, the following limitations should be taken into account. First, the effect of forced
migration on child cognitive development may be influenced by the duration of migration (temporary, permanent, and
returned migrants), the type of migration (rural to urban, urban to rural, and so on), and the urban-rural sampling structure.
On our part, we were unable to assess this due to the unavailability of such information in the YLS. Also, comparisons
of child cognition must be done at the sending places, particularly if the levels of socioeconomic development between
the sending and receiving places are very different. Again, we were unable to split the sample according to their current
migration status due to the unavailability of such information. Therefore, the negative effect of forced migration on child
cognition in our study may be due to the socioeconomic differentials between the migrant and non-migrant households.
However, the results obtained from PSM analysis also support the findings of the multivariate regression analysis.
Consequently, estimates obtained from the multivariate regression analysis can be safely taken as the actual effect of
forced migration on child cognition.
Second, we could not control the respondent/household characteristics prior to the forced migration due to the
unavailability of data. We did, nevertheless, control the following variables: respondent’s height, schooling of the
respondent, respondent’s age at birth of child and schooling of the head of household. We added respondent’s height as
a way of capturing genetic factors. We included the educational attainments of the respondent and the household head as
a proxy for wealth (that is, household economic status prior to migration). Although, a number of previous studies have
reported a high level of correlation between education and economic status, we must acknowledge the fact that all the
variations in wealth may not be captured by education and, thus, some care is required while interpreting the findings of
our study.
Third, the magnitude of the effect of forced migration on child cognitive well-being may be lower than expected due to
mortality selection among children from migrant households. Some previous studies have reported that forced migration
is significantly associated with higher infant and child mortality (Avogo et al., 2010). However, the effect of mortality
selection on child cognitive development should be minimal due to the fact that only very few deaths occurred betwwen
Wave 1 and Wave 3 in the YLS sample. Fourth, the observed differences in the cognitive well-being of migrant and non-
migrant children may be due to a potential selection bias resulting from attrition between the first and third waves of the
YLS. The attrition rate between waves 1 and 2 was about 3% and between waves 2 and 3 about 1% (Barnett et al., 2012).
Dercon et al., (2008) found limited evidence of attrition bias in the YLS and argued that the attrition in the YLS samples
were highly unlikely to cause a bias in research inferences.
Despite these limitations, our study has some strength. First, a large cohort of children was included in the analysis,
representing children from a wide range of family backgrounds. Second, the YLS is the only large-scale available dataset
that provides information on forced migration, social support, and child cognitive well-being in India. Third, the YLS uses
a child-focused mixed sampling approach, allowing for an examination of the complex interrelationship between forced
migration, social support, and child cognitive development in India. Fourth, the study included both rural and urban areas,
representing a range of regions, policy context, and living conditions that reflect the ethnic, geographical and religious
diversity of the country. Another key strength of this study is that the information on forced migration pertained to the
period between pregnancy of the index child and attaining one year of age. Some studies have reported that the first 1,000
days of the child (including the duration of pregnancy) is a very crucial period for child cognitive well-being at later ages
(Black et al., 2013). However, the majority of the previous studies have been unable to control for forced migration in this
crucial period in their analyses (Flores et al., 2009; Rossi, 2008). Lastly, our study came out with some findings that may
either lead to the formulation of new policies or may lead to the strengthening of the existing policies and programmes.
According to the United Nations High Commissioner for Refugees, the total number of displaced persons worldwide

International Journal of Population Studies | 2017, Volume 3, Issue 2 25