Sampling is studying a manageable subset to draw conclusions about an entire population you can’t measure fully. The quality of those conclusions depends entirely on how representative the sample is.
| Method | How it works |
|---|---|
| Simple random | Every member has an equal chance |
| Stratified | Split into groups, sample each proportionally |
| Systematic | Take every nth member from a list |
| Cluster | Randomly pick whole groups, sample within |
| Convenience | Whoever’s easy to reach (prone to bias) |
A good sample mirrors the population. Random methods guard against bias; convenience sampling often skews results. Stratified sampling ensures key subgroups are represented in the right proportions. Larger samples reduce random error, but a large biased sample is still misleading — representativeness matters more than size alone.
Frequently asked questions
Why sample instead of measuring everyone? It’s faster and cheaper, and often the whole population is inaccessible.
Most reliable method? Random-based methods; stratified when subgroups matter.
Does a bigger sample fix bias? No — bias must be addressed by good method, not size.
Always ask who a sample left out. A poll of only daytime landline users, or a survey only motivated customers bother to finish, can be large yet badly skewed. The missing voices, not the sample size, are usually where bias hides.