The Main Sampling Methods
You can do research that involves larger or smaller groups of people. If a group is large (the research results, in this case, will be more valid), you cannot collect data from every individual because it is rather time and effort-consuming. So, you need to select a sample. It consists of a chosen group of people who will take part in your research.
You have to decide how to choose this group of people to make your sample representative. So, you need to use one of the most appropriate sampling methods. There are two types of them - probability and non-probability sampling.
- Probability sampling is random. It allows for making stronger statistical conclusions about the entire group.
- Non-probability sampling is not random. It can be based on specific criteria that are most suitable in your case, for example, convenience or age.
No matter what sampling method you choose, you have to explain your choice in the methodology section of your academic paper.
The Difference Between a Population and a Sample
You have to distinguish clearly between a population and a sample, as well as define the target population.
- The population is the whole group of people you have chosen as the subject of your research. Your aim is to make accurate conclusions about the population, so you need sampling.
- The sample is a smaller group of people belonging to your target population. You will make conclusions on the basis of the data collected immediately from them.
You can define the population by using different criteria, such as their location, levels of income or education, age, social status, and many other features. The population is usually too broad to involve all its members in the research. Or it can also be too narrow to provide valid results. For example, it may include all the people inhabiting the vast region, or it may consist of all the customers of the company, patients with the same health problems, or students of a certain course from the whole country.
Define your population carefully. It has to meet the purpose, aims, and objectives of your research project. It is always difficult to involve all the members of the population because it may be geographically dispersed, demographically varied, or inaccessible. So, you need to choose a representative sample. You have to think about the sampling frame and sample size before making your choice.
What Is a Sampling Frame?
A sampling frame includes all the individuals from the population and excludes everyone who does not belong to it. It would be great if the sampling frame could involve the whole target population.
You want to research buying preferences of the customers who use the same online store. Your population is all current customers of 830 buyers. Your sampling frame comes from the database of this online store that lists the personal data of all the buyers.
What Is a Sample Size?
You can include various numbers of individuals in your sample. It depends on different factors like the population’s size and variability or the research design you have chosen for your project. You can use specific sample size calculators or formulas designed for statistical analysis.
How to Use Probability Sampling Methods
When you apply probability sampling, you can randomly select any member of the population. Such a method is often used in quantitative research projects. These techniques can produce the most valid results that are fully representative of the entire population. The four main types of probability samples are the following:
- simple random sample;
- systematic sample;
- stratified sample;
- cluster sample.
Using Simple Random Samples
Every member of the target population in a simple random sample can have equal chances to be selected. So, the sampling frame involves the entire population. You can utilize random number generators and other techniques to make simple random sampling. All of them are completely random and based on chance.
You are selecting a simple random sample of 50 high-school students in Green Town High School. There are 350 high-school students there, so you assign the numbers from 1 to 350 to all of them. Then, you select 50 needed participants with the help of a random number generator.
Use of Systematic Sampling
This technique is quite similar to the previous one. Though, it is easier to apply. All the population members also have numbers, but they are not selected randomly. You choose participants at regular intervals.
You list all the high-school students in alphabetical order. You choose the starting point from the first 10 numbers, for instance, number 4. Then, you count to 7 and pick up every 7th person on the list - 11, 18, 25, etc. You do it until you get 50 students needed for your project.
However, the sample may be skewed if there is a hidden pattern in it. For instance, if the database groups are divided by classes, there is a risk that you get students from the same class only, or girls will prevail over boys or vice versa. So, the sample is skewed in this case, so check which people in the list your numbering will involve before the start of sampling.
What Is Stratified Sampling?
Your target population may have subpopulations that differ by some important criteria. So, you need to identify these subpopulations to get more accurate conclusions at the end. Ensure that every subpopulation is equally represented in the sample.
- Divide the entire population into specific subgroups or strata related to one important characteristic, for instance, gender, age, income brackets, or interests.
- Consider the overall proportions of the population and calculate how many people to choose from each subgroup.
- Then, you can use random or systematic sampling to pick up participants from each subgroup.
Among all the 350 high-school students you are interested in, there are 280 boys and 70 girls. You have to ensure that you will keep the gender balance in your sample, so you form two strata - males and females. Then, you can utilize random sampling for each stratum and choose 28 boys and 7 girls, so your representative sample will consist of 35 individuals.
How to Use Cluster Sampling
You have to divide the population into subgroups here too. However, every subgroup needs the characteristics belonging to the entire sample. You may choose the entire subgroup instead of choosing individuals from it. If the time and place allow, you may take all the people from every sampled cluster. It must not be a large cluster. If it is, you need to apply one of the techniques described above. It is also known as multistage sampling.
This technique works well for large or dispersed populations. Though, the risk of error in such samples may be bigger because the differences between clusters may be substantial. So, you cannot guarantee that the chosen cluster is entirely representative of the target population.
You want to choose a sample of high-school students from 8 schools across the state (all of them are with approximately the same number of students in the class). You cannot travel to all the schools to collect data, so you pick out three schools. They will be your clusters.
How to Use Non-Probability Sampling Methods
Participants are not chosen randomly in a non-probability sample. You will not include everyone in a group. This type of sampling is simpler and cheaper, though there is a risk of sampling bias. Your conclusions can be limited because the inferences about the entire population are weaker. You have to make your sample representative of the population, but you can face some challenges.
These sampling techniques are mostly used in qualitative and explanatory research. These types of research designs do not require testing hypotheses about the whole population. You can just develop an initial view of a smaller population under research. There are also four techniques here:
- convenience sample;
- snowball sample;
- purposive sample;
- voluntary response sample.
Using Convenience Sampling
This type of sample involves people who are the most accessible. It is easy and cheap, and you can collect initial data conveniently. However, you cannot ensure that the sample is representative of the entire population. So, you cannot obtain highly generalizable results.
You want to research opinions on the residents’ satisfaction with municipal services in your city. You create a survey on the topic. However, since you are short of time and costs, you ask people in your residential area or in your block of flats to complete this survey. You involve only one part of the population, so this sample is not representative of the whole population of the city.
What Is Snowball Sampling?
You can ask other participants to involve respondents in your research project if the population is hard to access. If you have contacts with some representatives of this population, you can ask them to recruit more participants.
You need to research the views of disabled people in your city about the inclusiveness facilities the municipality provides. However, you do not know too many disabled people, just one person who is a volunteer in some associations. This person agrees to help you and puts you in contact with other disabled people in the area.
How to Use Purposive Sampling?
This type of sampling is often called judgment sampling. You can use your expertise as a researcher to choose the sample that will be the most suitable for your purposes. This technique is helpful and commonly used in qualitative research. You need to get extended information about the unique phenomenon or topic you research. You will not have to collect statistical data for this, or the population you are researching is too specific or too small. To make your purposive sampling effective, you need to develop clear criteria and rationale. They should be both inclusion and exclusion criteria.
Your purpose is to learn more about the experiences of people suffering from dyslexia. So, you need to select the sample purposefully, including only people with dyslexia in it. The number of participants in this sample cannot be extended because there are not so many people with this problem.
Voluntary Response Sampling and How to Use It
This type of sampling is similar to a convenience technique. The sample is easy to access. However, you do not need to choose participants as a researcher, but you find people who volunteer to respond to your survey published online.
Such response samples can be biased because volunteers are people with certain qualities, and other members of the population of your interest may not have these qualities.
You create and send out an online survey on customer support service satisfaction to all the customers of a big online store. Many of them have completed it, so you obtain insight into their feeling about the company. However, you cannot be sure that these answers express the overall opinions of all the clients because those who answered the questionnaire are always active in this sort of expressing opinions, or they have strong views of the store’s services based on their experiences, while others are more indifferent or not engaged so much. So, you cannot consider the received responses representative of all the customers.
Final Thoughts
The most widely used sampling methods can be probability and non-probability ones. You can choose the technique that you find the most appropriate for your research project.
However, remember that all the techniques may have their strengths and limitations. So, if you want to get the most unbiased, representative, and valid results, think about the combination of two techniques to ensure the research outcomes' validity.