Page 377 - Invited Paper Session (IPS) - Volume 2

IPS280 VEUN, Thy et al.
Total Without further With further stratification
Province Name number of stratification within within province
households province
LHK CDH GGH
Kampong Thom 144,539 1,055 300 360 570
Kampot 134,871 768 210 270 450
Kandal 237,056 1,515 390 450 810
Koh Kong 25,308 3,428 420 450 570
Kratie 68,234 1,192 330 360 570
Mondul Kiri 13,150 1,745 300 390 450
Phnom Penh 267,500 2,817 660 960 1,170
Preah Vihear 44,368 3,270 420 660 660
Prey Veng 243,701 1,117 300 390 630
Pursat 91,099 1,106 270 300 420
Rattak Kiri 32,095 2,456 480 540 840
Siem Reap 186,393 2,522 420 540 840
Preah Sihaknouk 43,179 3,217 390 450 480
Stung Treng 23,703 1,923 270 390 330
Svay Rieng 124,193 856 210 270 390
Takeo 194,113 955 270 330 570
Otdar Meanchey 46,341 1,357 300 390 540
Kep 8,568 516 150 150 150
Pailin 15,650 3,039 300 390 480
Thbong Khmum 171,980 1,352 330 450 780
Total 2,983,350 42,515 8,340 10,290 14,670

Among the three levels of required precision, the level of 1% CV required
too large sample size in all methods which would be unaffordable to conduct
due to budget constraints. The existing survey has ever been conducting with
up to 12,096 households only in the biggest sample year. Having considered
both the acceptable levels of precision and the cost of conducting it, only four
among the nine sets of samples from the said stratification with the lowest
sample sizes are finally proposed for the sampling design of household survey
in Cambodia with provincial domains. These four sets of samples were the
samples from the LHK and CDH methods with the 3 and 5% levels of required
precision. These four sets of sampled villages were selected by the systematic
probability proportional to size (sys-PPS) sampling scheme using the number
of households in each village as the size measure. These selected samples were
assessed by using the sample data on the employment rate and adult literacy
rate from the 2014 CSES. Table 3 illustrated about the results of assessments
on the performance of the two sets of samples from the LHK and CDH
methods with 5% CV. The assessment indicated that the two sets of samples
performed well within the acceptable levels of required precision. LHK5%
produced the estimate with 100% probabilities that the estimated CVs will fall
within less than 5% in 23 out of 25 provinces and capital for employment rate,

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