Sampling, the procedure of selecting a sample from a population of interest, plays a key role in research both in quantitative and qualitative data analysis. A representative sample provides the basis for making inference about a population, and the inferences based on a true sample have more generalizability about the world we live in. But, getting a representative sample from a population of interest is a real struggle in the practical world.
Sampling biases: A sample fails to represent a target population when it suffers from biases such as coverage bias and nonresponse bias. “Coverage bias occurs when the members of the sampling frame are systematically different from the target population in way that influences the result of the study” Remler & Van Ryzin, (2011). A sample having a good coverage may have very low response rate. The response rate is a product of contact rate and cooperation rate i.e. response rate = contact rate × cooperation rate. The nonresponse bias occurs when a significant number of members of a target sample refuse to respond and they have influence on the results.
Non-probability sampling: Though probability sampling is considered gold standard for generalizability, for many practical reasons, researchers choose non-probability sampling for many of their studies. The major forms of non-probability sampling include voluntary sampling, convenience sampling, and purposive sampling.
Voluntary sampling refers to the participation of members in a study responding voluntarily to an open call. Again, the concern of this kind of sampling is nonresponse bias called volunteer bias because the volunteers may differ from a more representative sample of the population. Convenience sampling refers to situations when researchers recruit participants from a natural gathering or from the people they have easy access to. Convenience sampling suffers from coverage bias because the people who are available to the researcher may not represent the target population of interest. Purposive sampling is a process of choosing participant with unique perspective or holding important roles to represent a theoretical category or considerations of a study. Snowball sampling is another type of sampling mostly used in qualitative studies where the participants are requested to refer the people they know for inclusion in the sample.
Probability sampling: Sampling technique that offers a chance or probability to each element of a population to be selected in a sample is called probability sampling. There are different types of probability sampling such as simple random sampling, systematic sampling, stratified sampling, and cluster sampling.
Simple random sampling is situations where each unit has equal chance or probability of selection. Systematic sampling is a technique of selecting every kth unit from a sample frame beginning at a random start point. Sometimes researcher divides the population into different groups called strata where the strata are mutually exclusive and exhaust entire population. Then, he chooses sample from each strata, which is called stratified sampling. Cluster sampling is another technique where the population is divided into different clusters sometimes following multiple stages, and finally one or more than one clusters are chosen as sample. The basic difference between stratified sampling and cluster sampling is that the strata are internally homogeneous and externally heterogeneous where clusters are internally heterogeneous and externally homogeneous. Random Digit Dialing (RDD) sampling has been used as a probability sampling technique since late 1960 where both listed and unlisted telephone numbers are given equal chance of being selected in a sample.
Sampling distribution: Sampling distribution is an important concept in probability sampling, which refers to the distribution of estimates (means) from many samples. The central limit theorem predicts that estimates of a large number of samples are distributed normally in a distribution, and the curve takes a bell shape, where mean, median and mode are equal. The standard deviation of sample means of the distribution is called standard error, and 68% of sampling distribution falls within +-1 standard error, 95% falls within +-2 standard errors, and 99.7% falls within +-3 standard errors.

-Remler & Van Ryzin (2011). Sampling. Research methods in practice: Strategies for description and causation

Leave a Reply

Blog authors are solely responsible for the content of the blogs listed in the directory. Neither the content of these blogs, nor the links to other web sites, are screened, approved, reviewed or endorsed by McGill University. The text and other material on these blogs are the opinion of the specific author and are not statements of advice, opinion, or information of McGill.