Sociological surveys of the population are often conducted among large groups of people. It is often erroneous to assume that the reliability of the results will be higher if the questions are answered by every member of the society. Due to the huge time, money and labor costs, such an examination is unacceptable. With an increase in the number of respondents, not only will the costs increase, but the risk of receiving incorrect data will also increase. From a practical point of view, many questionnaires and coders will reduce the likelihood of reliable control of their actions. Such a survey is called continuous.
In sociology, a non-continuous study, or a selective method, is most often used. Its results can be extended to a large set of people, which is called the general.
Definition and meaning of sampling method
Sampling method is a quantitative way of selecting a part of the studied units from the total mass, while the results of the survey will apply to each individual who did not participate in this.
Sampling is both a subject of scientific research and an academic discipline. It serves as a means of obtaining reliable information aboutgeneral population and helps to evaluate all its parameters. The conditions for selecting units subsequently affect the statistical analysis of the results. If the sampling procedures are poorly implemented, the use of even the most reliable methods of processing the collected information will be useless.
Key concepts of choice theory
The general population is the relationship of units, in relation to which the conclusions of a sample study are formulated. It can be residents of one country, a specific locality, the work team of an enterprise, etc.
The sample (or sample) is part of the general population, which was selected using special methods and criteria. For example, statistical criteria are taken into account in the formation process.
The number of individuals included in a given set is called its volume. But it can be expressed not only by the number of people, but also by polling stations, settlements, that is, definitely large units that include observation units. But this is already a multistage sample.
The sampling unit is the constituent parts of the general population, they can be either directly observation units (single-stage sampling) or larger formations.
A big role in obtaining reliable research results using the sampling method is such a property as the representativeness of the selection. That is, the part of the general population that became respondents,should fully reproduce all its characteristics. Any deviation is considered a mistake.
Steps in applying the sampling method
Each empirical sociological research consists of stages. In the case of a selective method, their order will be arranged as follows:
- Creating a sample design: the population is established, selection procedures are characterized, volumes.
- Project implementation: in the course of collecting sociological information, the questionnaires perform tasks indicating the method of selecting respondents.
- Detecting and correcting representativeness errors.
Types of samples in sociology
After determining the general population, the researcher proceeds to sampling procedures. They can be divided into two types (criteria):
- The role of probability laws in sampling.
- Number of stages of selection.
If the first criterion is applied, then the method of random sampling and non-random selection are distinguished. Based on the latter, it can be argued that the sample can be single-stage and multi-stage.
Types of samples are directly reflected not only in the stages of preparation and conduct of the study, but also in its results. Before giving preference to one of them, you should understand the content of the concepts.
The definition of "random" in everyday use has received a completely opposite meaning than in mathematics. Such selection is carried out according to strict rules, it is not allowedno deviation from them, since it is important to ensure that each unit of the general population has the same chance of being included in the sample. If these conditions are not met, this probability will be different.
In turn, the random sample is subdivided into:
- simple;
- mechanical (systematic);
- nested (serial, cluster);
- stratified (typical or regional).
Simple type content
A simple sampling method is carried out using a table of random numbers. Initially, the sample size is determined; a complete list of numbered respondents included in the general population is created. Special tables contained in mathematical and statistical publications are used for selection. Any other than them are prohibited. If the sample size is a three-digit number, then the number of each sampling unit should be three-digit, namely from 001 to 790. The last number indicates the total number of people. The study will involve those people who have been assigned a number in the specified range, found in the table.
Systematic type content
Systematic selection is based on calculations. An alphabetical list of all elements of the general population is preliminarily compiled, the step is set, and only then - the sample size. The formula for the step is as follows:
N: n, where N is the population and n is the sample.
For example, 150,000: 5,000=30. So eachthe thirtieth person will be selected to participate in the survey.
Socket type entity
A clustered sample is used when the population of people under study consists of small natural groups. In this case, it should be noted that the list number of such nests is determined at the first step. Using a table of random numbers, a selection is made and a continuous survey of all respondents in each selected nest is conducted. Moreover, the more of them took part in the study, the smaller the average sampling error. However, it is possible to use such a technique provided that the studied nests have a similar feature.
The Essence of Stratified Choice
A stratified sample differs from the previous ones in that on the eve of the selection, the general population is divided into strata, that is, homogeneous parts that have a common feature. For example, the level of education, electoral preferences, the level of satisfaction with various aspects of life. The simplest option is to separate the subjects by sex and age. In principle, it is necessary to carry out the selection in such a way that a number of persons proportional to the total number is singled out from each stratum.
The sample size in this case may be smaller than in a situation with random selection, but the representativeness will be higher. It should be recognized that stratified sampling will be the most costly in financial and informational terms, and nested sampling will be the most profitable in this regard.
Non-random quota sampling
There is also a quota sample. It is the only type of non-random selection that has a mathematical justification. The quota sample is formed from units that must be represented by proportions and correspond to the general population. In this form, purposeful distribution of features is carried out. If opinions and assessments of people are among the characteristics studied, then gender, age, and education of respondents are often quotas.
In a sociological study, two methods of selection are also distinguished: repeated and non-repeated. In the first case, the selected unit after the survey is returned to the general population in order to continue to participate in the selection. In the second option, respondents are sorted, which increases the chances of the rest of the population being selected.
Sociologist G. A. Churchill developed the following rule: the sample size should strive to provide at least 100 observations for the primary and 20-50 for the secondary classification component. It should be borne in mind that some of the respondents included in the sample, for various reasons, may not take part in the survey or refuse it at all.
Methods for determining sample size
The following methods are applicable in sociological research:
1. Arbitrary, that is, the sample size is determined within 5-10% of the general population.
2. The traditional calculation method is based on conducting regular surveys, for example, once a year covering 600.2000 or 2,500 respondents.
3. Statistical - is to establish the reliability of information. Statistics as a science does not develop in isolation. The subjects and areas of her research are actively involved in other related fields: technical, economic and humanitarian. Thus, its methods are used in sociology, in preparing for surveys and, in particular, in determining sample sizes. Statistics as a science has an extensive methodological base.
4. Expensive, in which the allowable amount of research expenses is set.
5. The sample size may be equal to the number of units of the general population, then the study will be continuous. This approach is applicable in small groups. For example, the workforce, students, etc.
It was previously established that a sample will be considered representative when its characteristics describe the properties of the general population with a minimum error.
An estimate of the sample size anticipates the final calculations of the number of units that will be selected from the population:
n=Npqt2: N∆2p + pqt 2, in which N is the number of units of the general population, p is the proportion of the studied trait (q=1 - p), t is the coefficient of correspondence of the confidence probability P (determined according to a special table), ∆ p – allowable error.
This is just one variation on how the sample size is calculated. The formula may change depending on the conditions and selected study criteria (for example, repeated or non-repeatedsample).
Sampling errors
Sociological surveys of the population are based on the use of one of the types of sampling considered above. However, in any case, the task of each researcher should be to assess the degree of accuracy of the obtained indicators, that is, it is necessary to determine how much they reflect the characteristics of the general population.
Sampling errors can be divided into random and non-random. The first type implies the deviation of the sample indicator from the general one, which can be expressed by the difference in their shares (average) and which is caused only by a non-continuous type of survey. And it is quite natural if this indicator decreases against the background of an increase in the number of respondents.
A systematic error is a deviation from the general indicator, also found as a result of subtracting the sample and the general share and arising from the inconsistency of the sampling methodology with the established rules.
These types of errors are included in the total sample error. In a study, only one sample can be taken from the population. The calculation of the maximum possible deviation of the sample indicator can be performed using a special formula. It is called the marginal sampling error. There is also such a thing as the mean sampling error. This is the standard deviation of the sample from the general share.
The a posteriori (post-experimental) type of error is also distinguished. It means the deviation of the indicators of the sample from the general share (average). It is calculated by comparing the generalindicator, information about which came from reliable sources, and a sample, which was established during the survey. Personnel departments of enterprises, state statistical bodies often act as reliable sources of information.
There is also an a priori error, which is also the deviation of the sample and general indicators, which can be expressed as the difference between their shares and can be calculated using a special formula.
The following mistakes are most often made in training research when selecting respondents for a survey:
1. Sample sets of groups belonging to different general populations. When they are used, statistical inferences are developed that apply to the entire sample. It is clear that this cannot be acceptable.
2. The organizational and financial capabilities of the researcher are not taken into account when types of samples are considered, and one of them is given preference.
3. Statistical criteria for the structure of the general population are not fully used to prevent sampling errors.
4. The requirements for the representativeness of the selection of respondents in the course of comparative studies are not taken into account.
5. The instructions for the interviewer should be adapted to the specific type of selection adopted.
The nature of respondents' participation in the study can be open or anonymous. This should be taken into account when forming the sample, since participants may drop out if they do not agree with the conditions.
