Defining a representative sample is a key success factor in any market research. For B2C market research, particularly the sample’s representativeness directly conditions the results’ accuracy. However, do not think a large sample is necessarily synonymous with quality. Smaller, well-stratified samples can yield much more accurate results than large random samples.
This article discusses key issues such as representativeness, the concept of size, and applications in quantitative (survey) and qualitative research.
Contact us for your survey projects
- What is a representative sample?
- How do you know if a sample is representative?
- The 2 types of criteria for a representative sample
- Example of a representative sample at the country level (socio-demographic criteria)
- How to determine the criteria for a representative sample?
- How to create a representative sample for qualitative research?
- Is a large sample size necessarily better than a smaller one?
What is a representative sample?
A representative sample is a sample that allows statistical inference to be made at the level of an entire population. In other words, a representative sample has the same characteristics as the target population and respects the statistical requirements to carry out an extrapolation.
A representative sample is mainly used in quantitative research: opinion polls, satisfaction surveys, etc. However, we will see in the following paragraphs that the question of representativeness can also arise in qualitative research.
How do you know if a sample is representative?
Every time you want to carry out research on a representative sample, you must ask yourself this essential question: “What are the criteria for the representativeness of my target population?”
This question is essential because the first dialogue between the customer and a market research firm like ours often looks like this:
- Customer: “I want market research on a representative sample.”
- IntoTheMinds: “representative of what?
- Customer: “of the entire population.”
- IntoTheMinds: “What are the criteria for a representative sample of your target population?
- Customer: ” I don’t know
Without this, you may interview 100,000 people, but your research will remain less accurate. Therefore, understanding the criteria that uniquely characterize your target population is essential. We will discuss the definition of representativeness criteria in the next two paragraphs.
Defining the representativeness criteria of your sample means anticipating the characteristics of your respondents that will play a major role in their future behavior.
The 2 types of criteria for a representative sample
The representativeness criteria are, therefore, the characteristics of your respondents that will influence their decisions. Generally speaking, there are two types: socio-demographic and behavioral.
Most people who conduct surveys swear by socio-demographic criteria: age, gender, and income, for example. These criteria play a role, but they are not necessarily preponderant. Basing the sample’s representativeness on socio-demographic criteria means that all people of the same age, sex, and income will behave similarly. This is, of course, reductive.
- Germany: Destatis
- Belgium: STATBEL
- Spain: INE
- France: INSEE
- Italy: Istat
- The Netherlands: CBS
- Switzerland: BFS
In an era of individualized behavior and hyper-personalized target marketing, behavioral criteria are undoubtedly the best way to stratify a representative sample.
Your target population’s habits, purchases, and possessions are probably much more predictive of their future decisions than their age or gender. Here are some examples.
- Buying a Porsche: Already owning or having owned a Porsche is the most important criterion (70%) for buying a new Porsche. Age or gender plays only a minor role.
- Purchase of a consumer product: If a product has already been purchased at least 7 times by a customer, you can assume that it has become a habit. Therefore, the past purchase of this item becomes the best predictor of a future purchase.
- A customer’s activity level in a savings bank: research we conducted in 2019 showed that socio-demographic criteria explained only 8% of a saver’s activity on their savings account.
- Use of an energy source: the location is often a determining factor in this type of research (connection to the gas network or not), as well as the type of dwelling.
Behavioral criteria are particularly important in notoriety research.
Example of a representative sample at the country level (socio-demographic criteria)
Country-wide surveys tend to be based on socio-demographic criteria. The table below shows the most frequently used criteria. The values for one country (France) are indicated.
|Criteria||The percentage to be reached||The example of France|
|Rural environment (<2000 inhabitants)||23%|
How to determine the criteria for a representative sample?
This brings us to the essential preparatory stage of defining the representativeness criteria. Defining the representativeness criteria for your sample means anticipating the characteristics of your respondents that will play a major role in their future behavior.
But how can you know in advance which criteria will play a role? There is little chance that the exact research you want to do has already been carried out by someone else and that the results are available. However, the behavior you want to assess has probably already been researched. Therefore, all you need to do is carry out documentary research to access this research. By reading them, you will be able to identify better the variables that have a proven influence and thus determine the representativeness criteria of your sample.
Example 1: market research for a new game console
In this first fictitious example, you want to research to assess the interest in a future high-performance game console. The research shows that high-performance consoles are used to play specific games (simulations, RPGs). Your survey will have to target people who already play these games to gather relevant opinions.
Your first representativeness criterion will therefore be behavioral: “owning a console.” The second criterion will be “play simulation or RPG games.”
Example 2: research on buying a foreign car
Buying a sustainable consumer good is a big project for market researchers like IntoTheMinds. In research on foreign cars, we found that socio-demographic criteria did not play a significant role. It was the brand of the current vehicle that was decisive.
How to create a representative sample for qualitative research?
The sample’s representativeness is often raised in the context of surveys (quantitative surveys). However, it can also arise in the case of qualitative research.
If you are conducting qualitative interviews or focus groups, your sample should cover the spectrum of socio-demographic profiles or behaviors likely to have an influence. Remember that qualitative research aims not to provide statistical certainty but to analyze the broadest range of profiles that fit the target.
The more profiling criteria you use, the larger your sample will be. We refer you to this page to help you determine the sample size. You will find all the information you need to use the calculator below.
Is a large sample size necessarily better than a smaller one?
Paradoxically, large sample size is not necessarily synonymous with accuracy. The fault lies in the representative nature of the sample. In this article, we explain how a Facebook survey of 250,000 people had a margin of error of 17% while a survey of 1000 people had a margin of error of 3 to 4%.
In conclusion, your sample choice determines your research’s quality. Defining a representative sample will assure reliable results and a minimized error rate, even with a small sample size. However, be careful not to go below certain minimum sample sizes:
- 100 respondents is the absolute minimum for a survey
- 15 respondents is a good starting point for qualitative research
Posted in Marketing.