Sampling is the process by which a group of observations that belong to a population is selected. This, in order to carry out a statistical study.

Sampling, in other words, is the procedure by which certain individuals belonging to a population that is being subjected to an analysis are taken.

Sampling is necessary due to the fact that populations can be too large and it is not feasible (economically and materially speaking) to collect data from all individuals.

The goal is for the sample to be representative. That is to say, that its indicators such as the average age, the average income, the percentage of men and women, among others, be the same, or very similar to that of the population.

Sampling types

The types of sampling can be distinguished based on different criteria. Thus, according to the technique to select the subgroup, the following can be differentiated:

Probability sampling

Observations are selected on the basis of randomness, that is, randomly. In this category we can find:

  • Simple random sampling: All individuals in the population have the same probability of being chosen as part of the sample. It has advantages, such as the fact that it is easy to carry out through computer systems. However, a complete listing of the entire population is required and, if the sample is very small, the selection may not be representative.
  • Systematic: An observation is chosen at random and, to select the rest of the sample, regular numerical intervals are used. That is, suppose I have a population of 10,000, and I randomly select observation 600, after which I can consider intervals of 30 observations. In this case, you would take the observations 600, 630, 660, 690, 720, 750, 780, and so on.
  • Stratified random: The population is divided into strata, which are groups that share common characteristics and are even more homogeneous than the population as a whole. Then, a sample is selected, either randomly or systematically, within each stratum. The objective is to achieve a representativeness of each stratum.
  • By conglomerates or clusters: It consists of creating groups smaller than the population, which reflect or share all the characteristics of this. Then we choose one of the clusters as a sample and analyze it in detail.
Stratified sampling

Non-probability sampling

The selection of the sample does not depend on the probability, but on the decision of the researchers. We can distinguish some subcategories:

  • Confidence method in available subjects: It consists in that the researcher will capture the subjects that are available to him. This, for example, in a geographical point at a certain time.
  • Opinion or intentional method: The researcher uses his judgment or criteria to choose who will participate as part of the sample. In other words, continuing with the previous example, the researcher could collect the sample at a specific place and time. But you could, given the research objectives, decide to include only those who are married and in their 20s and 30s.
  • Causal or incidental: The researcher directly selects the individuals who will be part of the sample. For example, to the students of a school. This, given that you have easy access to them.
  • Snowball: It consists in that, after finding the first subject (or first subjects) in the sample, the researcher asks him (or them) for help to identify other individuals with those same characteristics. It is a technique used when it is difficult to locate a specific group due to the handling of sensitive data, for example, illegal immigrants.
  • By quotas: The researcher, taking into account the composition of the population, and dividing by groups or strata, will make a proportional selection of the sample. For example, let's imagine that in the population there are 40% of people under 25 years of age, 35% of people between 25 and 50 years of age, and 25% of individuals over 50 years of age. So, a sample of 4,000 people would have 1,600 subjects under the age of 25, 1,400 between the ages of 25 and 50, and 1,000 adults aged 50 and over. It should be noted that the individuals who will cover each installment will be selected by some non-probabilistic method, that is, any of the techniques explained above.

Likewise, it should be noted that sampling can be simple, if done only once; double, when two samples are captured (the second can be used if the first does not yield definitive results); or multiple (it is similar to double, but with more than two samples).

Tags:  opinion comparisons economic-dictionary 

Interesting Articles