model name |
definition |
characteristics |
Simple random sample |
he entire population without any classification and order.
The sample was completely random. |
Each sample unit has the same probability of being chosen. Each sample unit is completely independent. There is no any correlation between them. Simple random sample is basic foundation for other kinds of sampling methods |
Systematic sample |
According to features or order, sample units in population were aligned and the arrangement appears graphic or table. Then they can be chosen at same distance or same interval. |
Sample units are uniformly distributed over the population. Sample size can be smaller than simple random sample. Using same distance or same interval to choose samples can be applied to the surveyed object with correlation to property rank and also can be applied to the surveyed object without correlation to property rank. |
Stratified random sample |
When population was heterogeneous considerably in feature, the population can be divided into strata as subpopulation. Stratum was relatively homogeneous . Then simple random sample units were carried out within strata.
It can increase level of precision |
|
Through stratification and classification process, it increases feature similarities among strata and makes relatively easy to get samples which were more representative of population. |
Spatial random sample |
Based on simple random sample and considering spatial autocorrelation, samples were drawn from spatial frame. |
It has smaller sample size than simple random sample. It works more efficiently as it avoids replicating samples with same feature. |
Spatial stratified sample |
Based on stratified random sample and considering spatial autocorrelation, samples were drawn from spatial frame. |
Considering spatial autocorrelation, it has smaller ample size than stratified random sample and has more precision than stratified random sample |
Sandwich sample |
Based on stratified random sample, It combines real application with reporting unit (customers expected to know) |
In terms of reporting unit and relationship among strata, it can directly get statistical results with high level of precision. |
