Sampling survey and some common methods

The field of sample survey involves how to get samples in an effective way. These surveys use questionnaires, and the design of questionnaires is very knowledgeable. It designs how to use words, the order of problems and the choice and combination of problems, etc. Involving psychology, sociology and other knowledge. The language of the problem should be adapted to the level of literacy of the respondents. One of the purposes of the survey design is to ensure that the sample is representative to the general to ensure the reliability of the subsequent inference. However each individual possible simple random sampling is an ideal condition.

The probability sampling method assumes that the probability that each individual appears in the sample is known. This probabilistic phase sampling method enables the data to be reasonably statistically inferred.

The non-probabilistic sampling method infers the data obtained from non-probabilistic sampling, and it relies on how the specific sampling scheme is designed and how it is implemented.

So what are the probability sampling methods:

1, the system sampling also becomes each n the name choice method (n-th name selection technique), this is first the overall number of each unit, and then randomly selected one of them as the starting point of sampling. The "distance" →n is determined based on the predetermined sample size, and after the start point is selected, the so-called equidistant sampling is usually performed by number at the starting point. For example, the starting point is 5, "distance" N = 10, then the following spot check objects are 15th, 25th, and so on. If the numbers are randomly selected, then this is equivalent to a simple random sampling.

2, stratified sampling is a simple random sampling of a variant, first to be studied by the general into a relatively similar or relatively homogeneous group of individuals, and then in each of the various types of simple random samples. Then the results obtained from the various types are aggregated and judged on the whole. The number of persons surveyed in each category is usually in proportion to that category, but may or may not be proportional or weighted for various considerations. (Weighted concept: When calculating the sum of several items, multiply the different coefficients for each, and the sum of these coefficients is usually 1)

3, cluster sampling is first divided into a number of groups, and stratified sampling is different in that the group is composed of non-similar or heterogeneous individuals, in single-stage cluster sampling, first (usually random) from these groups to extract a few groups, and then in these extracted groups of individuals in a comprehensive investigation. In the polar cluster sampling, a few groups are extracted from these groups first (usually randomly), and then a simple random sampling is done for the individuals in these extracted groups. Applicable to regional sampling, such as the investigation of villages in one county, it is clear that the situation in these villages is not very different, otherwise it will increase the error. The main application is the regional sampling, the group is divided by the region.

4, Multi-level sampling in large groups, often after the extraction of a number of groups, and then in which a number of subgroups, and even in subgroups to extract subgroups, and so on. This can be used at every level of different sampling methods, so more complex, also known as multi-level mixed sampling.

What are the non-probabilistic sampling methods:

1. The objective of sampling is to select the object subjectively by the researcher. Then the sample is much dependent on the prior knowledge.

2, Convenient sampling is usually used for initial assessment. For example, in order to investigate the views of tourists, may choose a different time and tourist attractions, feel free to stop the visitors to investigate. This may seem random, but it's not.

3, judging sampling by experience to determine the selection of samples, usually convenient sampling extension. This, for example, studies the situation in each county, and the researchers only sample in one county, then the county is representative.

4, the fixed sampling first determine the various types and proportions (similar to stratified sampling), and then use convenient sampling or judging sampling to select the number of individuals required by proportion.

5, Snowball sampling for the sample characteristics of interest more rare cases, such as drug addicts, you want to investigate a drug addicts related to the person, and then this person will introduce you to find the relevant person, may produce greater error. A method that relies on one goal to recommend another.

6, self-choice is to let individuals voluntarily participate in the investigation.

In general, the scientific data sampling method is many, but often is a combination of various sampling methods, because both the accuracy and the consideration of convenience, feasibility and so on.

Sampling survey and some common methods