Next Generation Myanmar > 53 Sampling The survey employed a multi-staged sampling plan. The 2014 Myanmar Population and Housing Census served as a sampling frame. In the first stage, Myanmar’s states/regions were used as strata, to which sample size was allocated proportionate to size of population. Small states/regions were boosted to ensure at least 100 respondents per region to improve accuracy in the region-level analysis. In the second, third and fourth stage, cluster sampling was applied. First, townships were selected randomly, within which village tracts and wards were randomly selected. Then, within these, household were selected using skip-step randomised household sampling from a starting point. In each household, a respondent was chosen from household members falling within the age bracket of the population, using last birthday as a random selection method. The appropriate sample size was calculated at 2,160 respondents (see Table 2). In each sampling point, a minimum of ten interviews were conducted. Interviewers conducted more than ten interviews in many points, thus increasing the sample size to 2,474 (accounted for in post-weighting). Table 2: Sample size by state/region/territory State/region/ territory Total youth population Youth population proportion (percentage) Calculated sample size Adjustment (boost) Adjusted sample Number of primary sampling units (PSUs) (10 respondents in each) Kachin 411,967 3.7 62.9 37 100 10 Kayah 66,003 0.6 10.2 90 100 10 Kayin 282,603 2.5 42.5 58 100 10 Chin 90,337 0.8 13.6 86 100 10 Sagaing 1,150,335 10.4 176.8 177 18 Tanintharyi 295,979 2.7 45.9 100 10 Bago 1,006,204 9.1 154.7 155 16 Magway 807,445 7.3 124.1 125 13 Mandalay 1,425,865 12.8 217.6 218 22 Mon 396,638 3.6 61.2 39 100 10 Rakhine 429,084 3.9 66.3 34 100 10 Yangon 1,866,067 16.8 285.6 286 29 Shan 1,323,980 11.9 202.3 203 20 Ayeyarwady 1,271,796 11.5 195.5 196 20 Nay Pyi Taw 283,067 2.5 42.5 58 100 10 Total 11,107,370 100.1 1,702 456 2,160 218 54