When prospects contact our market research agency to conduct a survey, they often come with the same request: that of a representative sample. This request doesn’t mean much. In this article, I’ll explain my take on survey representativeness and how to approach it.
Entrust us with your survey project
If you only have 30 seconds
- A sample can only be representative concerning well-defined criteria
- Representativeness criteria must be identified in advance, based on their influence on the researched behavior. In the case of market research, this means identifying the criteria that influence purchasing behavior.
- The most used representativeness criteria in B2C are:
- Age
- Gender
- level of education
- revenue
- location
- In B2B, it’s more difficult to build a sample because a company’s behavior depends on its processes and the people who work there. It’s, therefore, important not to multiply the criteria used to select your sample and to make do with a smaller sample.
- In B2B, the most used criteria for selecting a sample are:
- business sector
- company size
- respondent function
- location
Statistical representativeness: what is it?
The principle of statistical representativeness is to infer from results obtained on a sample. In other words, a statistically representative result can be extrapolated to the entire population under research. In surveys, the idea is to understand the behavior of an entire group by questioning only a limited number of people belonging to that group.
Without going into too much technical detail, it naturally follows that the larger the sample, the smaller the margin of error. You’ll have noticed that I’ve used the conditional tense, which is only sometimes the case. As I explained here, surveys on social networks are of extremely inferior quality despite their large sample size.
The question, therefore, arises as to what the word “representativeness” means in the context of market research.
Before discussing a representative sample, you must identify the criteria influencing buying behavior.
The importance of a representative sample in market research
There’s no doubt that representativeness is a central concept in all market research. The primary purpose of conducting such research is to reassure you about the potential of a product or service. So, the results had better be reliable.
Prospects contacting our firm often discuss representative samples without understanding what that means. They think “representativeness” is the magic criterion that makes the research reliable. The essential question they should ask is “representative about what? ”
The representativeness of a sample only makes sense if the representativeness criteria themselves make sense in the context of market research. Allow me to reason through the absurd. Imagine that men only buy a product. Why would it make sense to build a gender-balanced sample? On the contrary, the sample must be 100% male to be representative of purchasing behavior.
The main message I’m getting across is that before we can talk about a representative sample, we must identify the criteria that influence buying behavior. This approach may seem paradoxical. How is it possible to identify the criteria even before conducting the research? That’s what I’ll discuss in the next paragraph.
Send us your market research project
BtoC surveys: how to manage representativeness?
B2C is the field where the representativeness of a sample is most regularly discussed. This may be due to the use of opinion surveys in elections. To obtain reliable results, market researchers need to draw up samples that reflect the population that will be voting. This is a particularly difficult exercise, and we see certain institutes’ errors in their estimates at every election. The reason is simple.
Representativeness through quotas
To conduct these surveys, the institutes use a quota method (for more information, see our article on sampling techniques). Quotas are defined according to criteria of age, gender, education, origin, political persuasion, etc. But the question is how these criteria allow us to infer choices. Will every 24-year-old white male who has conducted a Master’s degree vote the same way?
To avoid this problem, you need to have an advanced understanding of what you’re researching. This is where desk research comes in handy.
The benefits of conducting desk research upstream
Desk research will enable you to identify, through other research, the factors most correlated with the behavior you seek to research. This preparatory phase is essential, as it will inform you of the criteria you must “control” in your sample.
We conduct B2C and B2B surveys throughout Europe. Contact us now!
BtoB surveys: is it possible to create a representative sample?
I want to start this paragraph by stressing the difficulty of B2B surveys compared to B2C. Professionals are much less available than individuals to answer questions, however interesting they may be. This inevitably impacts the size of the samples that can be gathered. So, it’s important to consider this before talking about representative samples.
The very concept of representativeness doesn’t seem at all relevant to B2B:
- companies are complex entities whose behaviors are rarely homogeneous
- companies’ internal processes determine, to a certain extent, the decisions they take
- individuals within companies also influence decision-making.
Suggested B2B approach
When we conduct B2B market research, we advise our customers to segment only a little. It’s better to focus on the “Big Picture” to identify patterns.
We advise to stay above 100 companies (N=100) regarding sample size. There may be particular cases where such a sample is unattainable because the population is too small. In such cases, a qualitative approach (semi-structured interviews) is preferable.
Criteria for building a B2B sample
We advise you to build a sample using a maximum of 2 of the following criteria:
- business sector
- size
- location
- respondent function
Conclusion
A survey offers a certain guarantee of reliability, provided that the respondents are correctly sampled. You need to understand the criteria for selecting respondents so that they are representative of the entire population.
Therefore, the term “representative sample” only makes sense if you already have an initial understanding of the factors likely to influence the behavior of those surveyed. If not, it’s a promising idea to explore these factors in advance through desk research.
Posted in Marketing.
