The representativeness of the sample is key in sampling.
Sampling is a procedure carried out to take samples that are considered representative of the average characteristics, quality or conditions of a larger group . The term is also used to name the choice of a small part of a statistical population with the purpose of making an inference about the value of some quality of the group.
The concept also allows reference to the technique used to carry out this type of selection. It is important to keep in mind that, in order for the properties recognized in a sample to be extrapolated to the whole, certain requirements must be met to reduce the margin of error .
Its characteristics
Sampling is used to choose a part of a statistical population . It is easier to study the sample (that is, the selected part) than the entire set; However, it is essential that the data obtained from the analysis of the sample be representative of the population in question. Otherwise, the inference will not be valid.
In practice, research generally cannot cover the entire universe of study. That is why sampling is used, the purpose of which is to make it possible to obtain reliable information about the population based on only one sample. Beyond precautions, inference always contemplates a margin of error that is measured according to probabilities.
In short, sampling is used to analyze the characteristics of a population by studying only part of it . The ability of researchers to carry out sampling correctly is key.
Many samples resort to randomness.
Sampling types
Depending on how the samples are selected, it is possible to differentiate between various types of sampling.
Systematic sampling involves making a list of the elements of a population, choosing one at random, and then continuing the selection according to a systematic interval. Cluster sampling , for its part, is based on a group of elements that are taken as units of analysis.
Random sampling , meanwhile, can be developed in two ways. Simple random sampling grants the same possibility of selection to all elements, while stratified random sampling (or stratified sampling ) contemplates the creation of strata before the election.
It should be noted that both systematic sampling, cluster sampling and random sampling are modes of probability sampling.
In non-probabilistic sampling , on the contrary, the choice of sample elements is the responsibility of the researcher, who makes use of his subjectivity and dispenses with chance. This type of sampling includes quota sampling (the population is divided into strata according to its characteristics and then a representative and proportional sample is taken from each one), snowball sampling (an individual is chosen, who then suggests another individual and so on) and purposive sampling (based on the convenience of the researcher to carry out the study).
A sample can arise from a finite population or an infinite population.
Research development
Researchers work with a statistical population : elements that share certain characteristics and that are of interest for an experiment, test or research. By carrying out a statistical analysis, you obtain information about the population in question.
As we already indicated, when the population is very large, sampling is used: a procedure or technique for selecting a subset. This subset is the sample , which represents the entire population.
The link between the sample size and the population size is known as the sampling fraction . The sample, on the other hand, is taken from the sample space (the set of possible samples that can be extracted through a certain sampling technique).
What the researcher does, in short, is to sample the population of interest and then analyze that subset. Thus, based on the data extracted from the sample, you can infer traits or properties of the entire population.
Sampling Examples
Suppose that an investigation aims to find out how many children under 8 years of age have their own mobile phone (cell phone) in a certain city . In this town there are ten primary schools, each with six classes attended by children of the age of interest for the research work.
For sampling, we choose to randomly select five children from each grade : this means that the statistical sample is made up of 300 children (there are six grades of interest in each of the ten schools, which means that sixty grades are considered; at take five children from each one, you reach 300). These children are asked whether or not they have a telephone: 60 say "yes" , while 240 say "no" . By statistical inference, it is stated that in the city in question 20% of children under 8 years of age have their own telephone .
Let's take the case of a company that manufactures product X industrially. To carry out quality control , you randomly select one item out of every ten you produce and examine it. This sampling allows you to analyze the conditions and properties of a certain percentage of your production. In this case, the selected products constitute the sample, while the total production is the statistical population.
Let's now think about a historian who is researching the descendants of aboriginal peoples in a country . In this framework, he manages to locate ten descendants and, in the absence of official records or institutionalized information, he asks these individuals to contact him with other men and women who also descend from the same ethnic group. This allows you to move forward with snowball sampling .