Random Sampling for Surveys and Experiments: A Beginner's Guide
By RithyVerse Studio · Updated October 12, 2026 · 3 min read
Whenever you want to learn something about a large group but cannot ask everybody, you take a sample. How you choose that sample decides whether your conclusions can be trusted. Random sampling is the most basic and most important technique for doing it well, and a random number generator makes it easy to do correctly.
What is random sampling?
In a simple random sample, every member of the population has the same chance of being chosen, and the choice of one member does not affect the choice of another. Because nobody decides who is included, personal preferences and convenience cannot bias the result.
Why not just pick people who are easy to reach?
Convenience samples, such as surveying friends or whoever walks past, tend to over-represent certain types of people. If you ask only the customers who are in your shop on a weekday morning, you hear mostly from people who can shop on weekday mornings. Random selection avoids this by giving everyone an equal opportunity.
How to draw a simple random sample
- Define the population. Decide exactly who or what you want to learn about, for example all students in a school or all invoices from last month.
- Build the list. Create a complete list of the population, called the sampling frame. Number every item from 1 to N.
- Choose the sample size. Decide how many items you need, for example 30.
- Draw numbers. Set the generator range to 1 to N, turn on No repeats, and draw numbers until you have the sample size you need.
- Select the matching items. The items with the drawn numbers form your sample.
You can do this with the Randomly generator and a spreadsheet in a few minutes.
Why no repeats matters
Sampling without replacement means each item can be chosen at most once. This is what most surveys require, because you cannot interview the same person twice. The No repeats option does exactly that: each number is drawn only once until the range is used up.
Other sampling methods
- Systematic sampling: choose a random starting point, then take every k-th item. It is simple but can be misleading if the list has a repeating pattern.
- Stratified sampling: split the population into groups (for example by year level), then sample randomly within each group to make sure all groups are represented.
- Cluster sampling: randomly select whole groups, such as classrooms, and survey everyone in them. It is cheaper but less precise.
How big should a sample be?
Size matters, but it is a separate question from randomness. A large biased sample is still biased. Larger random samples reduce the margin of error, but the improvement slows down: going from 100 to 400 responses helps a lot, whereas going from 1,000 to 1,300 helps much less. For a serious study, use a sample-size calculator or consult a statistician.
Common mistakes
- Redrawing until the sample "looks right". This undoes the point of randomness.
- Using an incomplete list, which leaves some people with no chance of being chosen.
- Ignoring non-response. If many chosen people do not reply, the final results may still be biased.
When a quick draw is enough
For informal tasks such as choosing which receipts to check or which students to interview, a simple random draw is plenty. For research that will be published, follow the standards of your field and document your method. Either way, the principle is the same: let chance, not preference, decide.
To understand what makes a generator fair in the first place, read How Random Number Generators Work.