Concept in Definition ABC

To study certain wide-ranging topics, scientists analyze a part of the entire object of their investigation. This part is a sample and the corresponding study is a sample. Thus, it is possible to carry out a sampling in all types of areas: analysis of the terrain, of solids, of sounds, demographic studies, blood analysis, urine, etc.

In any case, the sampling carried out aims to be a significant part of what is studied, such that the results obtained allow conclusions to be drawn from the set of what is study.

Types of sampling in the study of the population

Statistically know the set of a population it is an extremely difficult and time consuming question. Let us think of a study related to the diseases of a country, on the eating habits or the consumption of a some product. This type of analysis cannot be performed through a survey at a general level, so that sampling becomes a toolessential.

There is no single modality. In fact, there are several types of sampling. On the one hand, there are those that are random or probabilistic. On the other, the non-probabilistic ones. The first is the most suitable procedure to choose the subset of individuals from a population, since This tool makes it possible to guarantee that the chosen sample is representative of the entire population.

In the non-probabilistic type, the choice of sampling does not depend on the probability, but is related to the causes of the investigation or the purpose of the sample. Therefore, both procedures they respond to different objectives. Probability sampling is rigorous and scientific, while non-probability sampling is better suited to homogeneous population studies.

All sampling implies a certain margin of error

Suppose we want to know how many Spaniards are smokers. To establish the exact percentage, two procedures could be used: survey the 47 million inhabitants or either select a certain quantity as a representative sample of the entire population, for example, a sample of 1000 people.

If a random sampling is carried out, it is possible that the people chosen are mostly smokers and, therefore, the sample would not be representative. Consequently, in every sampling there is a margin statistical error.

To reduce the margin of error, experts advise choosing a suitable sample size. Of course, the larger the sample, the smaller the margin of error. In this sense, when the universe studied is large, it is necessary to use equally large population samples.

Photos: Fotolia - Andrey Popov / Alewka