How big should a good sample be?

We want to make a manual to guide our students (secondary education) in making a thesis. If they want to draw up a survey for their thesis, how large should their sample be? What guidelines do we give them?

Asker: Anna, 52 years old


Everything here depends on what you want to research with your surveys. To be able to make some reliable statements, I would certainly try to provide at least 30 respondents. What is more important is the way the sample is chosen. It is important to clearly define what and who you want to research.

If you want to investigate generalities, it is very important to take the sample as randomly as possible. This means that there must be enough variation between the respondents. For example, there must be a good mix based on gender, ethnicity, age, profession, etc. In this way, certain respondents are not over-/under-represented. In other words, your sample should look similar (in composition) to the population.

Suppose you really want to focus on a certain target group, then it is of course the intention to conduct the surveys effectively with that target group. But here too a good mix of gender, age, etc. is desirable. Then you can compare those groups again.

In general, I also think it is important to point out that they cannot generalize their results to, for example, Flanders, or Ghent, or … Even taking a representative sample of just one school is already a real task, but it could still be possible if the respondents are chosen entirely at random (and therefore not only respondents from their own social network). In this way, I think that in most cases the students are limited to statements about the sample itself.

Answered by

drs. Lennart Van Eycken


How big should a good sample be?

university of Ghent


Recent Articles

Related Stories