Page 60 - IJPS-8-2

International Journal of
Population Studies Projecting sex ratio at birth in Pakistan

in the modeling. As of now, no evidence of biased sampling of variability given the observation value (Brown, 1998).
has been reported for any of these data sources. With their On the provincial level, the SRB observations have a wider
own objectives of the survey, these data sources provide a range with greater uncertainties because of smaller sample
wealth of information, and there is no supporting evidence sizes. The smallest provincial SRB observation for DHS
to choose one at the cost of the others. Hence, we use all 2012 – 2013 is 0.917 in Islamabad (ICT) in year 2002 and
these sources in our model estimation and projection. the biggest value is 1.416 in Balochistan in year 1989. The
average sampling error of all the provincial observations
Figure 1 illustrates the large uncertainty in SRB for DHS 2012 – 2013 is 0.087, corresponding to 8% of CV.
observations on national and provincial levels. It shows Similarly, for Pakistan DHS 2017 – 2018, the observed
the observed SRB from Pakistan 2012 to 2013 and 2017 SRB on national and provincial levels ranges from 0.914
to 2018 DHSs on the national and provincial levels. In the to 1.178 and from 0.891 to 1.272, respectively. The average
data series Pakistan DHS 2012 – 2013, the national SRB sampling errors are 0.087 and 0.139 for national and
observations range from 0.935 in the year 1989 to 1.195 provincial data, respectively.
in the year 2003. The average sampling error associated
with these observations is 0.073, equivalent to 7% of the 2.2. Methods
coefficient of variation (CV; refers to the ratio of sampling The model performance and predictive power were assessed
error to the value of observations, indicating the extent by an out-of-sample validation exercise (leaving out recent

A

B

Figure 1. SRB data in (A) 2012 – 2013 and (B) 2017 – 2018 DHS on national and provincial levels
Note: The SRB observations (dots) from the same area (province or the whole of Pakistan) are connected with line segments of the same color. Shaded areas
around the observation series represent the sampling errors in the series (quantified by twice the sampling standard errors).

Volume 8 Issue 2 (2022) 54 https://doi.org/10.36922/ijps.v8i2.332