Every day we hear concepts of statistics without having the slightest idea of what they mean, and some of them affect us directly. One of the statistical terms that concerns the population is sampling. We explain what it means and the fundamental types of sampling that exist. Is easier than it looks like.
- Discover the 15 types of data analysis that exist.
What is sampling
Broadly speaking, sampling is the statistical technique for selecting a sample from a statistical population . We could also explain it in the following way: we can study a small group of the population to extract information or conclusions (demographic data, consumption habits, etc.), we need that small group of people to be representative of the entire population. The method to choose this group, so that it is representative for the rest of the population, is sampling.
The RAE offers the following definition:
“Action of choosing representative samples of the quality or average conditions of a whole”.
When is sampling used?
To know a series of variables or data about the population we can analyze the population as a whole. However, there are times when it is not possible to carry out a complete census , either because the population set is too large or because we do not have the necessary means and the data we obtain could be incomplete or full of errors.
- A good way to manage large amounts of information is Big Data.
A good way to measure population variables is to choose a sample that represents the entire set. In this way, the statistical study or analysis will be more precise, cheaper and will take much less time. At this point, we must ask ourselves two questions:
- Which individuals should we take as a reference
- How many individuals will be part of the sample?
The 8 types of sampling that exist
From the two previous questions, a range of sampling strategies opens up. Each of these types can be used in a specific case and has a number of advantages and disadvantages. We analyze them all.
1. Random Sampling
First, we list all random types. As can be deduced from the name, all individuals in the population have the same probability of being chosen for a sample, regardless of their peculiarities. It is done by generating random numbers, usually through computer systems. It is usually a more expensive and slow method.
1.1 Simple random sampling
This is the simplest method of the entire list. A number is assigned to each individual of the total population and through a random process , as many individuals as necessary are chosen to study the sample.
1.2 Systematic Random Sampling
As in the previous case, we assign a number to all individuals in the population. We choose one of them and the rest of the individuals on the list are added following an order with respect to the first (every x numbers or a multiple x). We must verify that the peculiarity that we study does not coincide with the periodicity.
1.3 Stratified random sampling
This method consists of two phases. In the first, we separate the total population according todifferent categories or criteria . These categories must be quite broad and homogeneous, for example sex or marital status. Subsequently, we randomly choose several individuals from each group until we reach the number necessary to complete the sample.
1.4 Random sampling by clusters
First, we divide the population into several groups or clusters according to their characteristics. Afterwards, each group is analyzed to determine which ones have more representativeness with respect to the total, while the others are discarded. Finally, we randomly select one individual from each group. Such groups often represent geographic areas.
1.5 Mixed random sampling
Sometimes necessarycombine two or more of the aforementioned sampling methods to obtain high representativeness.
2. Non-probabilistic sampling
The second large group of strategies are non-probabilistic, those in which the individuals in the sample are chosen following certain criteria. Chosen groups are less likely to be representative of the whole, but are generally cheaper methods.
2.1 Sampling of convenience
In this case chance does not intervene, but rather it is the criterion of the researcher himselfwhich determines which individuals can be part of the sample and will be representative of the population. Therefore, the degree of representativeness can never be calculated exactly. It is rare and should be considered when there is no better strategy.
sampling Also called accidental sampling, it bears certain similarities to stratified random sampling, but eliminates the random component. First we divide the total into several groups of individuals with common characteristics and then the first ones that meet these characteristics are chosen . An example would be a survey of Barcelona residents who are between 25 and 30 years old.
2.3 Snowball sampling
First, some individuals are chosen following certain criteria. As if it were a snowball sliding down the slope, these individuals lead us to others , following a chain until reaching the total sample. It is used with very small or marginal groups of individuals.
- Learn about the characteristics of the 16 types of research.
Casal, J., & Mateu, E. (2003). Sampling types. Rev. Epidem. Med Prev, 1(1), 3-7.
Cochran, WG, & Diaz, EC (1980). Sampling techniques (No. 04; HA31. 2, C6 1980.). Mexico:: Continental Editorial Company.