What’s it: Sampling is the process of selecting a subset of the population to use for research. A population can be things or people.
In this article, I am specifying this topic with market research.
In market research, the researcher must establish criteria and then select the number of respondents required. Surveying everyone in the market or population is impractical. Hence, sampling emerged.
If done correctly, the conclusions drawn from the sample can be applied to larger groups with measurable errors. Using samples, of course, saves costs.
Types of sampling
Broadly speaking, two types of sampling are:
- Random sampling
- Non-random sampling
In random sampling, each sample has an equal probability of being selected. The randomly selected sample is unbiased in representing the population.
There are several random sampling techniques, including simple random sampling, systematic random sampling, stratified random sampling and cluster random sampling, and multi-stage random sampling.
Simple random sampling
Simple random sampling is usually done with computer-generated random numbers. The random sample is based on the odds that each element has the same weight of being selected. Random samples can be used for both quantitative and qualitative research.
Systematic random sampling
This method starts with a random number, and then each N number after that becomes part of the sample population. For example, suppose there is a total population of 100. You decide to look at 10% of the population as a sample. In that case, you then generate random numbers from one to one hundred and choose one as a starting point.
Let’s say that the random number chosen is 45. Then, you choose every 10 numbers after or before number 45 until you reach the sample size. So, in this case, the selected samples are 45, 55, 65, 75, 85, 95, 05, 15, 25, and 35.
Stratified random sampling
Stratified sampling is used to ensure the sample size is representative of the broader population. For example, in a population of 100, there are 60 men and 40 women. If 10 people were to be sampled, it would consist of 6 men and 4 men. A randomized or systematic approach was then used to select 6 men from 60 people and 4 women from 40 people.
Cluster random sampling
Cluster random sampling is a sampling technique by which many groups of people are created from a population. Members in one group have homogeneous characteristics. However, characteristics should be heterogeneous for members between groups.
In this method, a simple random sample is made from each group in the population.
For example, if you want to do research to assess students’ performance studying business in Indonesia. You cannot possibly involve all business students in Indonesia.
As an alternative, you can use cluster sampling. You can group universities from each region into one cluster. This cluster then defines all the second-year student population in Indonesia.
Furthermore, using either simple random sampling or systematic random sampling, several clusters can be selected for the research study. Next, using simple or systematic sampling, you choose students from each group.
Multi-stage random sampling
Multi-stage random sampling divides a large population into several stages to make the sampling process more practical. Typically, it uses a combination of stratified sampling or cluster sampling and simple random sampling.
Let’s say you want to know which subjects school children prefer in Indonesia. Population lists – a list of all school children in Indonesia – are nearly impossible to come by, so you can’t research the entire population.
Instead, you can divide the population into provinces and select sample representative provinces randomly. For the next stage, you can take a simple random sample of schools from each province. Finally, you can do a simple random sampling of students at school to get a sample.
Non-random sampling, or non-probability sampling, is sampling in which certain elements of a population will have a higher chance of being chosen than others. Respondents are selected with some of your subjectivity. You feel they represent the population they are trying to survey.
Types of non-probability sampling include quota sampling, convenience sampling, judgmental sampling, snowball sampling, and self-selection.
In quota sampling, the quota is determined by the characteristics of the population. For example, a researcher might want to interview 100 buyers. If 10% of the shopping population are elderly, you can then interview 10 elderly buyers.
It is where you select a person who is willing to stop and be interviewed. This is different from simple random sampling because you didn’t choose respondents randomly beforehand.
Convenience sampling is often used as a pre-test questionnaire to determine whether the questionnaire questions are understandable and reliable.
Judgmental sampling occurs when you use expert opinion to determine what makes a representative sample. This sampling is usually more biased. For example, suppose someone chooses to interview buyers in the middle of the workday before lunchtime. In that case, the business person will likely not be well represented.
This is a type of sampling in which the current respondent recommends the next few respondents.
This method is beneficial for identifying particular groups of people to interview. For example, when interviewing drug dealers, you may not know who the dealers are.
One way to get the next respondent is to ask the current respondent which dealers are still in their network.
In this sampling, respondents make a decision whether to participate in the survey or not.