Survey ID Number
CHN_2001_WVS-W4_v01_M
Title
World Values Survey 2001, Wave 4
Sampling Procedure
Sample size: 1000
The sample is a representative national sample of China containing 40 county/city sample units to collect individual level data of, from a political cultural perspective, the values and attitudes currently held by Chinese citizens. With considerations of representativeness, feasibility, and budgetary constrains, it was decided this project would draw a subsidiary probability sample out of a master sample that RCCC created based on its previous national survey on environmental awareness of the general public in China conducted in 1998. The Environmental Awareness Survey, which was used as a master sample, was a national survey conducted through out the entire country. The target population was the same as the one defined for this survey. Through the stratification, the proportionally allocated multi-stage PPS (probability proportional to size) technique was employed in order to obtain the self-weighted household samples. There were different stages in the sampling procedure: Counties and county-level cities are taken as primary sampling units (PSUs). Family households are the basic sampling unit. Demographic data at all levels was obtained from The Demographic Data for Chinese Cities and Counties, 1997, published by the State Bureau of Statistics.
Nation wide, there were 2,860 county-level units for the first stage sampling (including 1,689 counties, 436 county-level cities, and 735 urban district--with administrative rank equivalent to
county--in large cities). The total households were 337,659,447. This was the base for establishing the sampling frames. Some readjustments: Taking into account of cost and accessibility, only the provincial capitals (Lhasa and Urumchi) and their surrounding areas in Tibet and Sinkiang were included in the sampling frame; in other remote western provinces, a few areas that are extremely hard to access were left out as well. After such readjustment the sampling frame then includes 2,708 county-level units, of which the total households are 322,002,173. Compared to the target population, there was a 5.3% reduction (152 units) in the first stage sampling units. However, since the population density in the remote areas of the western provinces is very low, the reduction counts merely 1.4% of the total households in the sampling frame. Geographical administrative divisions of China were regarded as the primary labels of stratification, that is, each province was treated as an independent stratum. Allocation of target sampling units among the sampling stages was designed as following: 135 PSUs out of the first sampling (county-level) units; 2 secondary sampling (townshiplevel) units in each of the PSUs; then 2 third sampling (village-level) units in each of the SSUs; 25 households in each of the third sampling units, on average. Based on the proportional stratification principle, sample allocation to strata was proportional to the size of each stratum, by an equal probability of f = .0042%. Within each stratum (province), sample sizes were calculated and allocated proportionally to each of the sampling stages. A self-weighted national sample thus was obtained.
Multi-stage PPS:
-The first stage: equidistance PPS was employed to draw the county sample.
-The second stage: in each of the chosen county-level units, a sampling frame was created based on the data of townships/ward and size measurement; then the equidistance PPS is employed to choose the township/streets sample.
-The third stage: a third sampling frame was obtained from each of the chosen township-level units (neighbourhoods, villages and size measurement), and, again, the equidistance PPS is employed to choose the village/neighbourhood sample.
-The fourth stage: in each of the chosen village/neighbourhood units, the official list of households registration was obtained; using the size measurement of this unit and the desired number of households to count the sampling distance, then households were selected according to the sampling interval. Since the household registration also listed all family members of each of the household, respondents were drawn randomly immediately after the household drawing. The WVS-China sample was drawn out of the above described master sample.
Some readjustments: Primarily because of the budgetary constrains of the WVS project, six remote provinces in the master sample were excluded. They were: Hainan, Tibet, Gansu, Qinghai, Ningxia, and Sinkiang. These provinces are all with very low population density, and all together they count 5.1% of the total population and 4.6% of total households of the country. After the adjustments, seven of the 139 county-level units of the master sample were removed. Therefore, the target 40 PSUs were to be drawn out of the remaining 132 units.
Sampling Stages:
-The first stage: 40 units were drawn from 132 county-level units of the master sample were removed. Therefore, the 40 PSUs were to be drawn out of the remaining 132 units.
-The second stage: one unit was chosen randomly out of the 2 original township-level units (SSUs) in each of the 40 selected PSUs.
-The third stage: one unit was chosen randomly out of the 2 original village-level units in each of the selected SSUs.
-The fourth stage: from each of the chosen village-level units, 35 households were drawn out of the household registration list with equidistance, along with one respondent in each selected household.
Remarks about sampling:
-Sample unit from office sampling: Housing