Choosing the correct sample size – that is the number of people you recruit to participate in your survey – is an essential part of running a survey and gathering key insights about the broader population of interest. Let’s look at an example of what that means.
Suppose you are at the bakery and see a chiffon cake topped with fruit and lemon frosting. You’d like to know what it tastes like – you’d like to sample it. To get a full picture of what the cake tastes like, the sample has to be a certain size. You can’t just take a pinch. With a pinch, all you may get is the cake part and miss the frosting and fruit.
It’s no different from a survey. In a survey, you recruit a sample of the population as a means to gather insight into the broader population. (The broader population is the population of people that matter to you e.g. customers, potential car buyers, millennials, U.S adults, etc). To do that, your sample has to be large enough.
Rule of thumb for sample sizes
Most of the time, you’ll be studying large populations (e.g. populations consisting of thousands or millions of individuals). It is common practice in market research to use 200-300 participants as a starting point for determining sample size for your study. With at least this many participants, you’ll have the ability to make inferences about the wider population of interest.
Two hundred survey participants are not always feasible, however. There may be financial or logistical hurdles standing in the way. In this case, 100 survey participants are sometimes accepted, but the lower number represents a compromise in the quality of your data.
If you just need a rough or “ballpark” view of the underlying population being studied, even fewer participants are needed. In this case, 30 to 60 survey participants are sufficient for doing investigational work and developing – rather than validating – a hypothesis.
At 200 survey participants, your data will be robust enough to base business decisions on. You’ll have the opportunity to derive conclusions from your data at the 95% statistical confidence level.
What if you need to be more precise than that? Well-known political polls that strive for a very high, academic-level precision use samples of about 1,000 individuals to get information about national attitudes and opinions.
Anytime your research goal involves detecting significant differences between customer segments, your total sample size must be set to accommodate the sample size requirements of each group. For example, if you want to understand gender differences, your sample should include at least 200 females and 200 males. That means a total sample size of 400.
Evaluate your needs
These rules of thumbs on sample size should serve most all your survey requirements. There are always unique situations, however, and sample sizes should also be evaluated to fit your specific needs. Recruiting more people to your survey means more costs, and especially if your target is difficult to reach, a minimum sample of 200 or even 100 survey participants may not be feasible or financially possible. You might also have a situation where your population is less than 100 individuals. What then? Obviously, the 200 minimum rule couldn’t apply. For more custom situations, sample size calculators are available, and these can help you evaluate the change in statistical precision for various population and sample sizes.