Random sampling includes five types: simple random sampling, systematic sampling, classified random sampling, cluster random sampling, and multi-stage random sampling. The following explains related definitions, examples, and scope of adaptation.
1. Simple random sampling
1.1 Definition:
1.2 Example:
- Direct sampling
- Lottery method or lottery method, all sampling units are numbered, and the numbers are written on the film to form a ball;
- Random number table method (randomness can be guaranteed);
1.3 Adaptation scope
2. Systematic sampling (Isometric random sampling)
2.1 Definition:
Based on a certain sampling distance, samples are drawn from the population.
- Number the population;
- Determine the segment distance, and use simple random sampling to determine the first individual number in the first segment;
- The remaining samples are drawn according to certain rules, and the samples are usually drawn according to the segment distance.
3. Stratified random sampling (type random sampling)
3.1 Definition:
- Divide the overall units into various types (or layers) according to certain standards;
- Determine the number of sample units drawn from each type according to the ratio of the number of units of each type to the total number of units;
- According to the random principle, samples are drawn from each type.
3.2 Example:
We want to understand the production and operation of 400 state-owned enterprises in a certain city, and decide to adopt random sampling method to select 20 enterprises as samples for investigation. The specific methods are as follows:
- Divide these 400 enterprises into three categories according to industry (also according to administrative division, profitability, scale, etc.), assuming 40 in the primary industry, 200 in the secondary industry, and 160 in the tertiary industry.
- Determine the number of sample units for each type of enterprise according to the proportion of each type of enterprise in the overall. Among them, the enterprises in the primary industry account for 10% of the total, and 2 enterprises should be sampled in proportion; according to the same method, 10 enterprises in the secondary industry should be sampled, and 8 enterprises in the tertiary industry should be sampled.
- Use simple random sampling or equidistant random sampling to select the above-mentioned sample sizes from various enterprises.
If a company has seven salespersons, a sample can be drawn from the customers of each salesperson based on the number of customers served by each salesperson.
3.3 Adaptation range
- Advantages:
It is suitable for survey subjects with a large number of overall units and large internal differences . Compared with simple random sampling and random sampling equidistant, in the same number of samples, it is pumping sampling error is small ;
small number of samples required sampling error is the same, it is required. - Disadvantages:
Must have more understanding of the overall situation of each unit, otherwise it is impossible to make a scientific classification. This is often difficult to achieve before actual investigations.
4. Cluster random sampling (collective random sampling)
4.1 Definition:
- Divide the sampling units in the sampling frame into many groups according to certain standards, and treat each group as a sampling unit;
- According to the random principle, select several groups from these groups as survey samples;
- Investigate all the sampling units in the above sample group.
4.2 Example:
Parts produced in one hour can represent parts produced in one week.
4.3 Adaptation scope:
If the quantity remains the same or there is no major change, this method is appropriate.
5. Multi-stage random sampling (multi-stage random sampling, segmented random sampling)
5.1 Definition:
Divide the sampling process into two or more stages; the
sampling steps are as follows:
(1) First, divide each unit of the survey population into several clusters according to certain signs, as the first-level unit of sampling, and then divide the first-level unit into several clusters Small clusters are used as the second-level unit of sampling. By analogy, it can be divided into third-level and fourth-level units.
(2) According to the principle of randomness, first select a number of units from the first-level unit as the first-level unit sample, and then select the second-level unit sample from the first-level unit sample, and so on, you can also draw the third-level unit sample Unit sample, fourth-level unit sample. The survey work to the second-level unit sample is two-stage random sampling; the third-level unit and the fourth-level unit sample are three-stage or four-stage random sampling.
5.2 Example:
5.3 Scope of adaptation
Combining the advantages of various random sampling methods, the requirements for understanding the overall situation of the survey are relatively low. Generally, sampling can be done only by understanding the composition of the next-level unit. The best sampling effect can be achieved with the minimum consumption of human, financial and material resources.
It is especially suitable for survey subjects with a large overall survey scope, many units, and complex situations.