Sampling Methods — Random, Stratified & Systematic

Sampling methods are the rules used to choose a sample from a population. In statistics, the method matters because a biased sample can give a misleading result before you do any calculations.

The three common methods are simple random sampling, stratified sampling, and systematic sampling. The right choice depends on the population, the list you have, and whether important groups need separate representation.

Sampling methods at a glance

Simple random sampling uses chance alone, so each member of the population has an equal chance of selection.

Stratified sampling splits the population into meaningful groups, called strata, and then samples randomly inside each group. Use it when those groups matter to the question.

Systematic sampling starts at a random point on a list and then takes every $k$th item. It is fast, but only works well if the list order does not contain a repeating pattern related to what you are measuring.

When to use each sampling method

Use simple random sampling when the population is already one mixed pool and you mainly want fairness through chance.

Use stratified sampling when some groups are important enough that a plain random sample might miss them or underrepresent them.

Use systematic sampling when you have a long ordered list and want a practical rule such as "take every 10th name." The condition matters: if the list order has a hidden cycle, the method can distort the result.

If you want a quick rule, ask this:

• Is the population basically one mixed group? Use simple random sampling.
• Do specific groups need reliable representation? Use stratified sampling.
• Do you have a long neutral list and need a fast method? Use systematic sampling.

Worked example: stratified sampling keeps key groups in the sample

Suppose a school wants to survey study habits. There are $100$ students: $60$ juniors and $40$ seniors. The school wants a sample of $20$ students.

With a simple random sample, the result might be close to the true split, but chance could still produce too many juniors or too many seniors.

With stratified sampling, the school keeps the same proportions in the sample:

$$20 \cdot \frac{60}{100} = 12$$ $$20 \cdot \frac{40}{100} = 8$$

So the sample includes $12$ juniors and $8$ seniors, chosen randomly within each year group. That makes sense here because year level could affect study habits, and the school wants both groups represented in the same proportions as the population.

Stratified sampling is not automatically better in every problem. It helps when the groups are meaningful and you care about representing each one well.

How systematic sampling works

Imagine the same school has a list of $100$ students and wants a sample of $10$. A common interval is

$$k = \frac{100}{10} = 10$$

So the school might choose a random starting point from $1$ to $10$, then take every $10$th student after that.

This is efficient, but it has a weakness. If the list is arranged in a repeating way, such as by class period or program, every $10$th choice could keep hitting the same type of student too often. In that case, the convenience of the method becomes a source of bias.

Common mistakes that bias a sample

Calling every sample random

A sample is not random just because the chooser did not plan carefully. Random sampling requires a chance-based rule.

Using stratified sampling without a real reason for the groups

The groups should matter to the question. If the strata are arbitrary, the extra complexity may not help.

Ignoring the order in systematic sampling

Systematic sampling is only safe when the list order does not create a harmful pattern relative to the interval.

Confusing a sample with the whole population

Even a good sample is still a sample. It gives an estimate, not perfect certainty.

Where sampling methods are used

Sampling methods appear in surveys, opinion polls, quality control, experiments, public health studies, and classroom data projects. In each case, the sample is chosen first and the analysis comes later.

That is why sampling belongs at the beginning of statistics, not the end. If the sample is weak, the averages, charts, and conclusions built from it can also be weak.

Try a similar problem

Take a population you know well, such as a class, a club, or a list of products. Choose one sampling method and justify it in one sentence. If you pick systematic sampling, give the interval and check for a repeating pattern. If you pick stratified sampling, explain why the groups matter before you calculate the sample sizes.