Representative Sample
Representative Sample — a group of objects that has all characteristics that are necessary for research. A representative sample is a part of the universal set, and their characteristics are similar. A good representative sample is a group of thirty persons where fifteen people are old, and the other fifteen are young.
Definition of Representative Sample
Representativeness is the most important quality of data that helps to build analytical models. It reflects the ability of the data to represent the dependencies and patterns of the subject area under study. That’s why it is necessary to have a representative sample.
Representativeness can be qualitative and quantitative. Quantitative representativeness shows whether the number of elements in a sample is enough to represent the universal set. Qualitative representativeness shows that all elements of the universal set are reflected in a sample. The metaphor of representative sample and universal set is meadow and bouquet. The universal set is a meadow with numerous flowers, and the representative sample is a bouquet with the widespread flowers.
A representative sample can be created based on the proper selection of sampling algorithms. The size of the sample doesn’t matter. The thing that matters is the presence of all necessary groups. The researcher should select the proper subgroup to do so.
There are several ways to violate the representativeness of the sample: when the wrong samples are taken and when too many (or few) samples are taken. The smaller the sample size, the less likely it is to be representative. The consequences of violating the representativeness of the sample are quite negative: incorrect conclusions of the study, the wasted budget, and financial losses due to the use of incorrect conclusions.
Criteria of Representative Sample
The representative sample should comply with a range of criteria to be considered valid. The researcher should take into account and observe the conditions under which the sample properly represents the general population, and in each specific case, establish with what confidence, and to which particular population, the results of the sample observation can be generalized. Here are the criteria for a representative sample.
Certainty and homogeneity of a representative sample and universal set. All samples should be equal in the necessary characteristics. The samples with different typological characteristics can’t be placed in one sample. In such cases, it is completely unclear to understand which properties and dependencies identified in the sample can be generalized.
No subjective influence. The fulfillment of this condition provides randomization. It means a random selection of options from the general population, where each unit of observation has the same probability of being included in the sample. Violation of this principle deprives the researcher of the right to generalize the results of a selective study to the general population and, thereby, deprives the selective study of its meaning.
No discrepancies between the same-name parameters. The sample should coincide well with the qualitative and quantitative parameters of the universal set. A researcher should create conditions that can represent the universal set perfectly. In every case, they should define with what confidence, and to which particular set, the results of sample observation can be generalized.
Methods of Representative Sample
Several methods are utilized to create the representative sample. They are used in statistics, finances, investing, marketing, pharmaceuticals, and other spheres.
Random sampling. Random sampling is a perfect way to create a representative sample. In this case, any member of the universal set can be part of the sample regardless of their characteristics. As a result, the sample would be a small copy of the universal set.
Stratification. Another effective method is stratification. According to this method, the major part of the population is divided into homogenous subgroups. Such a subgroup is called strata. Then an equal number of group members are selected from each layer.
Systematic sampling. The third method is systematic sampling. In this case, participants or study items are selected from a random starting point to begin with. The sampling is then carried out at fixed periodic intervals.
Stratified random sampling. This method consists of two stages. First, the total number is divided into subgroups. Next, elements are randomly selected from each layer, using a simple random sampling method. For example, it can be a public opinion poll for different segments of the population by a selection of several opinions from each segment of the population.
These samples are then combined to form a stratified random sample. Unlike simple random sampling, stratified random sampling ensures that the necessary data layers will be represented in the sample. The evaluation of parameters obtained from a stratified sample has greater accuracy and less variability or variance.
