Page 373 - Invited Paper Session (IPS) - Volume 2

IPS280 VEUN, Thy et al.
are increasingly becoming important. Cambodia Socio-Economic Survey
(CSES), which has been conducted regularly every year from 2007 to 2017, and
every two-year afterwards, provides such estimates only at national and
regional levels. This limitation posts a lot of challenges to policy makers and
other data users in their policy studies and preparations, and their related
statistics compilations at their desired granulated levels. Because of the
unavailability of these data, the current estimates of such statistics are based
dominantly on the benchmarks from the national estimates which somehow
distort the actual representation of individual provinces. Aside from these
statistics, there are many other economic and social indicators for the 2030
Sustainable Development Goals which are needed to measure and compile at
provincial level using data from this household survey. Being able to do so,
the study on the development of alternative sampling designs for the existing
household surveys which can provide the reliable estimates of such indicators
at a more granulated level, specifically at provincial level, is required. This study
shall contribute insights on the construction of a new sampling design for
household surveys in Cambodia with provincial domains. This study aims to
search for sampling techniques and statistical methods applicable on the kind
of population characteristics of Cambodia.
The sampling design of the existing CSES from the 2014 to 2017 were done
based on the stratified three-stage sampling design. In the stratification, the
frame of all villages in Cambodia was stratified into 38 strata frames (19
province groups and urban-rural). Among the 19 province groups, 14 of which
correspond to individual provinces and 5 of which correspond to grouped
provinces as a combination of two adjacent provinces. From each frame,
implicit stratification was also carried out before the selection of sample
villages by sorting the frame according to their geographical orders. The
sample size of urban villages was determined for 30% of the total sample size.
For each stratum the sample size was then allocated proportionally according
to the number of villages of the stratum among the whole urban or rural areas
of the country.
With goal of providing provincial estimates, this study aimed to compare
different methods in stratifying the population based on the distribution of
the target variables of CSES. Section 2 provides description of the data used
and the different stratification approaches compared in this study. Section 3
discusses the results of the study, whereas Section 4 provides some
conclusions based on these results.

2. Methodology
The most updated list of all villages in Cambodia from the 2008 population
census with some updates on the administrative and other related changes
was used in this study. This has served as the frame for all household related

360 | I S I W S C 2 0 1 9