There are various sampling types and sampling techniques; these techniques have been categorized into three major categories for the ease of use.
Each of the above mentioned sampling types have several sampling techniques in them.
Random sampling is also called as probability sampling. In random sampling design every element of the population has an equal
and independent chance of being included in the sample. This is by far the best sampling design if the researcher knows each element in the population and he is able to choose any element from it. It improves the validity of the study because each element included in the sample have been selected randomly without any bias or personal preferences. There are several types of random sampling designs:
Simple random sampling
Simple random sampling as the name suggests is the most basic type of random sampling. It fulfills all the criteria of random sampling. Each element in the simple random sampling design is chosen using either table of random numbers or through draw. The researcher identifies each element in the population, list them down. He decides about the number of elements in the sample and pick out the required sample randomly from the population. Simple random sampling is easier to use if the population size is small, in case of large population like a community or a city population the researcher has to use other type of random sampling technique.
Stratified random sampling
When the population is large and heterogeneous the researcher cannot use simple random sampling. There is another sampling design that is called as stratified random sampling and it is best suited to situations where the population has greater variation. The researcher identifies the population in this sampling design. He then decides about the number of units in the sample. The researcher makes strata of the population. The reason for dividing the population in strata is that the population is too diverse and cannot be treated in a simple manner. The researcher can make strata according to any special characteristic of the population. From each strata the researcher chooses units in the sample. The number of sample from each strata can be chosen in two ways. One the researcher can take proportionate units from each stratum, the other way is to take disproportionate, equal elements, from each stratum. It depends on the type of research and its requirements.
Systematic random sampling
In systematic probability random sampling the researcher selects every kth element in the population. The value of k can be taken by dividing the number of units in the population by the number of units in the sample. For example there are 40 units in the population and the sample size is taken to be 20 so 40/20 is equal to 2 and hence every 2nd unit in the population will be taken as a sample. In household surveys this technique is most commonly used for sampling.
In stratified random sampling the researcher divides the population in strata but in cluster sampling the researcher identifies clusters or groups in the population. Units are selected from each cluster and taken in the sample. These clusters or groups are usually naturally found in the population and the researcher does not divide the population himself.
Non-random sampling designs
Non-random sampling designs are also known as non-probability sampling designs. These samples are not taken on the principles of probability. It does not mean, however, that these samples are not representable, valid or generalizable to the whole population. There are times when the researcher cannot take random sample from the population and he is forced to select a non-random sample. For example, the research is conducted on the behavior of transsexual people in a community. Taking a random sample is difficult as not all the units or members of that population will be ready to share their views or fill questionnaires. The researcher has to ask them and if they will be willing he can only then take observations or interviews. In psychology, social sciences and behavioral sciences there is always a posed risk of not getting a random sample and hence non-random sampling techniques has to be used. There are different types of sampling designs in this type of sampling:
In quota sampling the investigator or the researcher decides about the number of samples to be taken and then he can freely choose any sample from the population. There is no distinction and he can choose any unit.
The investigator in this type of sampling selects the units from the population according to his own judgement. The reason might be that the investigator thinks certain elements in the population to be more fit for the survey than other.
In snowball sampling the investigator selects one element in the population randomly or non randomly to ask questions. The investigator asks that individual to identify another element of the population that can be taken in the sample.
In convenience sampling the researcher or the investigator selects samples according to his convenience. He selects sample that are easily available or more easy to ask questions. These are only few of the sampling types in research and there can be many more depending on the type of study.