Definition of the research sample
Simple random sampling is the most common method. In this case we choose pieces of the sample independently, by the same probability. Each member of the population has the same chance to be included in the sample. First of all, we need the list of the complete population, then we pick names randomly, or assemble the required size of sample using a random number table. One type of this sampling is the one without replacement, where one item can not be included in the sample twice.
Stratified sampling is a modification of the simple- and the systematic sampling, which may further enhance the representatitivity of the sample. It enables equal inclusion of items from the population from homogenous subgroups that we wish to examine.
The simplest – although not the only one – method of stratified sampling is when we choose a sample size based on a stratification feature from each group of the population, according to a predetermined ratio, with simple random sampling. The stratification depends on what information we have about the population.
systematic sampling we pick for example every fifth member of the population to be included in the sample (this is the sampling interval: the distance between the items).
To avoid certain bias in the sample we may choose the first number randomly, in this case from the first of 5 items, and then each fifth would be included in the sample. The sampling ratio is the ratio of the items included in the sample to the population; here: 1/5.
The threat of the method occurs when the sampling frame consists of groups according to some sort of ranking, or is periodical.
Concentrated sampling further decreases the role of chance and enhances the responsibility of the researcher who carries out the sampling. In this case individuals are weighted, and only those are included who are considered to be opinion leaders.
In the case of arbitrary (expert) sampling the researcher can define which individuals he includes in the sample. The only limitation to influence the decision is the required number of the sample size.
Cluster sampling means that we sample the groups of items, and in the next step we sample again within these groups. Thus we prepare a list and apply the sampling multiple times, applying selection.
Quota sampling is quite similar to stratified sampling, as we apply auxiliary information in this case as well. The content of the domestic areas are defined according to residents and grouped by locations, and a quota (list) of required respondents is being assembled according to the most important features. The researcher who carries out the sampling – in the knowledge of the quota – must examine all individuals who meet the designated features as long as the predetermined quota is fulfilled.
Sampling with replacement results in a sample where all items were returned to the pool before an item is selected. To execute the method first we need a list of all members of the population, from which we randomely choose the first piece of the sample. Examples for this method are hard to find in practice, but this is what happens during an oral exam when the lecturer returns each topic to the pool after each presentation, thus it may occur that one topic is discussed more than once.
Nested sampling aims to avoid the negative feature of simple random sampling when the listing of the population’s members is problematic. Thus in this case we concur to practice as we do not need information about the actual members to be examined , only about certain important features (criteria or grouping) of them. Based on these clearly understood features we create primary sampling units, and pick the items of the sample from this pool. Then we examine each item of the thus – randomly – assembled sample (without having knowledge of their existence prior to the sampling).
