As you probably know, sampling is a crucial step in survey research. It is a reliable method for conducting surveys and relevant market research. But what is sampling in concrete terms? What method should you use to get the results you want? And how can you go about selecting your sample? You will find all the answers to your questions in our article !
What is sampling?
Sampling consists of selecting a group of individuals representative of the reference population for a market research study. In other words, since it is impossible to interview everyone, you need to think about a realistic sample in order to generalize the results obtained afterwards. This technique makes it possible to conduct large-scale surveys using a sample of the population to replace the whole.
The vocabulary to master for good sampling
In order to be comfortable with this method, it is important to master the basic concepts and vocabulary such as
- The reference population : refers to the entire population you are interested in. This is the entire population of interest as defined in the survey objectives.
- The sampling frame : is the official and available list of your chosen reference population. Directories are not always accurate; there may be a margin between the actual reference population and the contact information you find.
- Coverage Error : the difference between the reference population and the sampling frame is the coverage error
- Sample : refers to the group of individuals selected to represent the reference population
- Sample unit : this refers to the basic unit of the survey, i.e., the people who respond to the survey
- The sampling error :refers to errors made during the selection of the sample. It is a reliable method, but a margin of error must always be considered.
What type of sampling to choose : random or non-random?
When conducting market research, there are several methods available to you. First, you have two options: random or non-random sampling.
- Random sampling : from the population studied, selection of a sample in which each individual has the same probability of being chosen. This method is related to chance.
- Non-random sampling : subjective selection of individuals to constitute the sample. Since non-random sampling does not require a complete sampling frame, it is a quick, easy and inexpensive way to obtain data.
When considering drawing conclusions and generalizing the information collected, it is advisable to use a random sampling method, which will tend to be less biased.
Let’s look at the different methods of these two types of sampling.
1. Random sampling methods 🔍
Simple random sampling
The pollster will select individuals at random, respecting their anonymity as well as equal treatment. Each person has the same probability of being selected, often they are assigned a number and drawn at random. This is a simple and inexpensive method to implement. It eliminates any potential bias in the sampling process.
However, because of its randomness, it can result in a group of respondents that is not representative of the reference population.
In this technique, the pollster begins his sample on a first sampling frame by listing the individuals. He begins by selecting a random starting point for the first sample member, to which he adds a fixed, regular interval. This periodic interval will then allow for consistency in the selection of individuals and the sample.
Example: If the size of the reference population is 8000 and you want to interview 500 people, you will have to divide the population by your sample, i.e. 8000/500 = 16.
You will then select the elements of the target population at regular intervals of 16.
The pollster subdivides the target population into homogeneous subgroups, called strata. These strata are defined according to one or more criteria (the variables of interest).
In each stratum, the pollster randomly selects individuals to obtain simple random sub-samples.
The objective of this method is to obtain a representative sample because all the individuals in a group have the same probability of being part of the sub-sample and the sample obtained is representative of the population with respect to the variable of interest.
Example: You are conducting a survey on the well-being of employees at work at X with 200 employees. The sample was obtained by randomly selecting 50 managers, 50 assistants, 50 alternates and 50 communicators. This will give you an overview of the different positions within this company.
Here, the pollster directly selects groups rather than individuals. These can be predefined groups such as people sharing the same postal code or belonging to the same company.
This sampling method can be conducted on two levels: randomly selecting the cluster and then randomly selecting the individuals within the cluster.
Example: Working in an airline company, you decide to conduct a customer satisfaction survey over the course of a day. You randomly select 10 flights for the day and interview all passengers on those flights.
Cluster sampling allows you to obtain a representative sample of the reference population if the clusters are similar and the individuals are heterogeneous.
2. Non-random sampling methods 🔍
The pollster will interview only those people he or she knows or who are available to complete the survey at a time T. This method is appropriate for an early or preliminary stage of a future, more accurate survey. However, if you want to obtain representative results on a large scale, you will need to use a more scientific method.
This method is very common. Here, the pollster will identify different criteria that will allow him to define quotas by category. This analysis can be based on characteristics such as
- demographic criteria
- use of a certain product
The size of the quotas is proportional to the relative size of each category within the population. This method ensures that the convenience samples reflect the structure of the actual population and its quotas.
The pollster uses his or her own judgment or that of another competent person to identify the people to be included in the sample. The selection criteria are subjective and practical.
Participants may be selected based on:
- Their knowledge
- Their understanding of the research question
- Or their objectives
This necessarily implies that some members of the population will be less likely to be selected than others.
This sample also presents significant risks of bias.
Here, the people recruited are invited to contact their friends and family to ask them to be part of the sample as well. These people are then invited to do the same around them, etc. This is a very useful method if you don’t have in-depth knowledge of the target population, and therefore little way to recruit many participants.
However, snowball sampling can often be biased because it promotes a homogeneous sample. Some groups will be more represented than others, and some rather isolated individuals will not be interviewed at all.
Note : As you can see, each sampling method has its advantages and disadvantages. To avoid errors as much as possible, or at least minimize them, it is important to know the possible risks and to choose the right sample size to survey.
You don’t know how to transcribe your survey ?
Noota is there for you ! This tool allows you to convert all your audio files into text in just a few minutes.
Then, create your own tags and classify your data to efficiently gather relevant insights. Feel free to check out our online site for more information, or contact us directly! We would be delighted to discuss with you.