1. Simple random sampling: Simple random sampling refers to extracting a part of observation units in the population in a completely random way (that is, the probability that each observation unit is selected into the sample is the same). The common method is to number all the observation units in the population first, and then draw some observation units from them to form samples by lottery, random number table or computer-generated random number.
Its advantages are simple and intuitive, and the calculation of mean (or rate) and its standard error is simple and convenient; The disadvantage is that when the group is large, it is difficult to number the individuals in the group one by one, and the samples taken are scattered, so it is difficult to organize the investigation.
2. Systematic sampling: systematic sampling is also called equidistant sampling or mechanical sampling, that is, all individuals in the group are sorted and numbered according to the characteristics unrelated to the research phenomenon; Then, according to the sample content, the sampling interval k is specified.
The advantages of systematic sampling are: easy to understand, simple and easy to operate; It is easy to get a sample evenly distributed in the population, and its sampling error is less than that of simple random sampling. Disadvantages are: the samples are scattered and it is difficult to organize the investigation; When the observation unit as a whole shows a periodic trend or a monotonous increasing (decreasing) trend, it is easy to produce bias.
3. Cluster sampling: Cluster sampling is to divide the population into k "groups", each group contains several observation units, and then randomly select k groups.
The advantages of cluster sampling are that it is convenient to organize investigation, save money and control the quality of investigation; The disadvantage is that the sampling error is greater than that of simple random sampling when the sample content is constant.
4. Stratified sampling: Stratified sampling is to divide all individuals in the population into several "layers" according to some characteristics that have great influence on the main research indicators, and then randomly select a certain number of observation units from each layer to form samples.
The advantage of stratified random sampling is that the sample is representative and the sampling error is small. After stratification, different sampling methods can be adopted for different layers according to specific conditions.
The characteristics of sampling survey are:
1. Sampling survey The survey samples selected from the population inferred from the survey are selected according to the principle of randomness. Because they are not influenced by any subjective intention, all units in the population have the possibility of being selected.
It can ensure the reasonable and even distribution of the selected survey samples in the population, and the possibility of tendentious deviation in the survey is extremely small, and the samples are very representative of the population.
Second, the sampling survey takes all the selected survey samples as "entrustment" to represent the whole, rather than using a randomly selected single unit to represent the whole, so that the survey samples are fully representative.
Three, the number of survey samples selected by sampling survey, according to the difference between the surveyed units and the allowable error of overall inference, determined by scientific calculation. Because there is a reliable guarantee in the number of survey samples, the samples will be very close to the overall reality.
Fourth, the sample error in the sampling survey, before the survey, can be calculated according to the overall differences in the number and units of samples, and the sample error is controlled within a certain range, so that the accuracy of the survey results is more certain.