by: Sherwin B. Ragos, PhD
When investigating the population is difficult if not impossible, it is appropriate to take a sample. That sample can be a representative of the population. Sampling is the process of taking samples from a population. It is cost-effective and convenient. That's why we do a lot of it in our everyday life. We'll have foretaste of the food we cook, we scan books or magazines that we're about to buy, we watch trailers of upcoming movies, and many more. In business statistics, we also conduct sampling to study a population of products, employees, customers, etc.
Sampling can be of two types: probability and non-probability. The choice between these two depends on the study to be conducted. This is crucial because a sample that misrepresents a population results to sampling error, and it could only alter the inference we would do about the population. The essence of probability sampling is randomness while non-probability is judgmental.
Probability Sampling Techniques
Simple random. This sampling is also known as "fishbowl" sampling method. In the selection process, all in the pool of items have equal chance of being selected. There are two kinds of simple random: with replacement and without replacement. If upon selection of items, the names or codes are removed from the sample frame, it is a simple random without replacement so that the items cannot be selected again. When upon selection of items, the selected names or codes are returned to the sample frame, the items have another chance of being selected again and it is a simple random with replacement.
Systematic Random Sampling. In this type of sampling, the items on the list are coded first. If the N (population) is a 2-digit number, the codes must also be 2-digit numbers. For instance, N=80 and you'll take n=10, the code of the first item is 01, second is 02... last item is 80. The first item of the sample is a randomly pre-selected number from 01 to 80. Then,
k = N/n
where: N = population, n = sample size
Assuming the first number is 12, add k=80/10 or 8. Hence, the second item is 20. The value 8 will be added to 20 and so on until the desired sample size is completed.
Stratified Random Sampling. In this sampling technique, the population is divided or grouped according to a certain 'strata' or classification. If we want to take 10 samples from 50 students using simple random, there is a possibility that all of them are males or females. If that happens, in either case, there will be no representative of the opposite sex. It won't be a good representative of the population. That's why we group them first according to sex and perform a simple random.
Cluster Sampling. This sampling technique resembles much like a stratified random sampling applied in a wider scope or geographical area.
Non-probability Sampling Techniques
Quota Sampling. The researcher, in this technique, sets a quota for the sample size. The researcher then fills-in the quota until the quota is reached or completed.
Purposive Sampling. This sampling technique has no randomness involved. The researcher sets a criteria and seeks for the items that meet the criteria.
Convenience Sampling. This is also known as the 'man on the street' sampling technique. If the researcher intends to know a prevailing opinion of the general public about an issue, he can administer a convenience sampling.
Snowball Sampling. The researcher selects a participant based on his judgment. Then, after taking information or data from the participant, he asks suggestion from the respondent about who he thinks can participate in the study until the desired sample size is completed.
