Statistical Sample – What it is, characteristics, types and population

We explain what a statistical sample is, its characteristics and what types exist. Also, what is a statistical population.

statistic sample
The statistical sample is isolated for evaluation and study purposes.

What is a statistical sample?

A statistical sample (or in contexts explicitly referred to statistics, a sample) is understood as a more or less representative subset of a statistical population, isolated from the rest for evaluation and study purposes. In other words, it is a fragment of the totality of elements to be studied, made up of a more manageable number of them, selected (ideally) at random.

The logic behind taking a statistical sample is that, given the right conditions, a very bulky set can be studied through smaller portions that are representative, that is, they are more or less proportional to the rest.

For example, if we want to study the universe of the millions of voters in a country, we must take a sample large enough to bring us, in a small group of a few hundred people, a reflection of the political opinions that exist in the population. whole. Thus, from a population of millions of individuals, we would study a sample of hundreds of them.

Said samples are obtained through different statistical techniques, which guarantee through different mechanisms an adequate randomness for the least possible bias in the selection, that is, the greatest possible objectivity that allows obtaining valid approximations to the statistical universe. If, on the contrary, a biased sample is obtained, the possible conclusions will be less reliable and therefore less useful.

Obviously, every sample is part of a population, so if you have several populations, you must also have several samples. Sampling is the process of obtaining a statistical sample and it is common in disciplines as different as demography, biology or politics.

Characteristics of a statistical sample

Broadly speaking, a statistical sample is characterized by the following:

  • Be part of a larger set, which is the statistical population or statistical universe, of which it is, ideally, representative.
  • It has a small number and therefore manageable of elements of statistical interest, compared to the entire population.
  • Is chosen at random and through different sampling techniques. It can be more or less reliable, depending on the latter.
  • Its size is the object of mathematical study, in order to guarantee the fair proportions so that it is representative of the total.

Statistical sample types

The statistical samples are classified, first, into two large groups: probabilistic and non-probabilistic, each with its own independent classification.

Probabilistic statistical samples. They are those that are chosen through more or less random methods, to guarantee the least intervention of the researcher’s criteria in the sample. In turn, they are classified into:

  • Simple random samples. The simplest of all are chosen absolutely at random from the population. This is the case, for example, of a national public opinion poll for which some citizens are elected by their document number.
  • Stratified samples. They are chosen randomly from among the different strata or levels of classification into which the population has previously been organized. For example, the sample can be chosen at random among the different age ranges of the population, thus obtaining a random but stratified sample.
  • Cluster samples. Similar to the stratified ones, they are chosen randomly from a previously determined set, but in this case these sets are not the result of the researcher’s criteria, but are given in a spontaneous, natural way. For example, a sample of the residents of a certain neighborhood, or of the workers of a certain building.

Non-probability statistical samples. They are those whose selection is not left to chance, but to certain search criteria of the researcher, due to limitations that prevent a larger sampling. Therefore, these types of samples are not really representative of the statistical universe studied, but they allow obtaining an approximation, endowed with a certain margin of error. These samples can be of the following types:

  • Intentional samples. Those that are chosen according to the researcher’s criteria, that is, taking those that he considers will give better results, as they are more representative. An example of this is when a journalist asks opinions of certain people that he has chosen beforehand.
  • Samples for convenience. Those that are chosen according to what is closer to hand, that is, limited to the immediate. This is what happens, for example, when a representative of a company offers its products to those who pass by.
  • Consecutive samples. Those that are part of a researcher’s journey, who goes from group to group, extracting the data to later constitute a whole. An example of this is the methods of approaching the public of certain sellers or promoters, in which they invite people to stop to listen to the virtues of the product: some do it and others do not, and later the seller changes area. In the end, you will gather all the data from the different areas you were in.
  • Quota samples. It is a combination of stratified samples and intentional samples, since the researcher chooses the people to interview according to their belonging (and representativeness) of a certain stratum or group determined in advance.

Statistical population

A statistical population differs from a statistical sample in that the latter is part of it, since a population is equivalent to the totality of the elements or individuals of interest for the investigation. In other words, the statistical population is the statistical universe: the whole, the entire mass of possible research elements.