Sampling Procedure
The 2008 Mongolian Reproductive Health Survey (RHS) has a nationally representative sample of 8382 households, in which 9402 women of reproductive age 15-49 years and a subsample of 3362 husbands were interviewed.
Complete national representation is key for any sample to give results that are truly generalizable to an entire population. The current RHS used the same sampling techniques as the 1998 and 2003 surveys to achieve national representation with minimal sampling errors. The survey was conducted using a two-stage sampling method, which gives an equal probability of selection of households. As this is the same method used in previous surveys, this means that the data are fully comparable with the RHS 1998 and RHS 2008 data sets. The sample frame was comprised of the listings of households prepared annually by bags and horoo across the country. Activities, such as improving household listing, designed for improving the quality of data of the first-stage sample frame were conducted in the fourth quarter of 2007 and the first quarter of 2008. Based on that the actual sampling was based on data from the first half of 2008.
It was determined from the experience of RHS 1998 and RHS 2003 and the similar surveys conducted in other countries that 25 to 30 households per cluster would provide an optimum representation; therefore, 30 households were selected for a cluster. It was decided select to 8,400 households (the lower limit was 7,560), which is 1.3 percent of all households in the country. Dividing this number by the 30 households by cluster gave the result of needing 280 primary sampling units (PSUs). Baghs and horoos were chosen as the primary sampling units (PSUs), resulting in a total of 1,676 PSUs. These PSUs were stratified implicitly by aimag and soum, and the selection of the 280 sample PSUs (or clusters) was done systematically with a random start and probability proportional to the number of registered households in each PSU. Households were then selected systematically with a random start within each PSU, using an interval directly proportional to the number of households in the PSU. The selected households were interviewed using the household questionnaire. All women between the ages of 15 and 49 (inclusive) who slept in the household's dwelling the night prior to the interview were eligible to be interviewed using the women's interview schedule. Three husbands out of every five married women were interviewed in each PSU using the husband's interview schedule.
Distribution of the RHS Household Sampling by Aimag, Mongolia 1998, 2003, 2008
-- Number of households in 1998, 2003, 2008 and Clusters
1) Arhangai, 300, 390, 330, 1-11
2) Bayan-Olgii, 200, 300, 270, 12-20
3) Bayanhongor, 250, 330, 270, 21-29
4) Bulgan, 175, 240, 210, 30-36
5) Gobi-Altai, 175, 240, 210, 37-43
6) Dornogobi, 125, 180, 180 ,44-49
7) Dornod, 200, 240, 270, 50-58
8) Dundgobi, 150, 180, 150, 59-63
9) Zavhan, 250, 300, 270, 64-72
10) Ovorhangai, 350, 420, 390, 73-85
11) Omnogobi, 125, 180, 180, 86-91
12) Suhbaatar, 150, 210, 180, 92-97
13) Selenge, 250, 330, 300, 98-107
14) Tov, 275, 330, 330, 108-118
15) Uvs, 249, 300, 240, 119-126
16) Hovd, 200, 300, 270, 127-135
17) Hovsgol, 325, 450, 420, 136-149
18) Hentii, 200, 270, 240, 150-157
19) Darhan-Uul, 250, 270, 300, 158-167
20) Ulaanbaatar, 1607, 2610, 3060, 168-269
21) Orhon, 149, 300, 300, 270-279
22) Govisumber, 50, 30, 30 ,280
Total number of households in 1998, 2003, and 2008 respectively 6005, 8400, 8400