07 / 12 · Learn from a sample
Who actually answered?
Separate the group you care about from the people you actually counted.
Builds on From 20% to 30%: what actually changed?
Go to practice ↓A question to keep in mind
If you only ask library visitors at noon, can their answers represent every reader?
The population is the whole group you want to understand. The sample is the part you actually observe. A sample can help, but how people were chosen matters.
Only asking at noon leaves out people who cannot visit then. Asking many more noon visitors does not automatically fix that gap. A random sample uses a defined chance-based selection process.
Work through an example
| Group | Count |
|---|---|
| Registered readers | 1000 |
| Noon visitors asked | 20 |
| Those 20 who want longer noon hours | 16 |
- Target: all 1,000 registered readers.
- Observed: only 20 visitors who were there at noon.
- 16/20 = 80% supported the change in this sample. That alone does not establish 80% support among all readers.
Your turn
0 / 3What is the clearest problem with using this survey to describe all registered readers?
Read solution · does not award completion
Readers who do not visit at noon are not included.
Keep the valid calculation, but limit the conclusion to what the selection process supports.
Match each part of the library example to its role.
Read solution · does not award completion
- All 1,000 registered readers → Whole group of interest
- The 20 noon visitors asked → Observed sample
- 16 of the 20 supported longer noon hours → Result within the sample
The target, the people observed, and the result among them are three distinct pieces.
A separate sample has 50 readers, of whom 30 prefer weekend hours. What percentage of this sample prefers weekend hours?
Read solution · does not award completion
- Sample percentage: 60 %
30/50 = 60%. This is the sample result. Whether it describes all readers depends on how the sample was selected.
Bring it back to your own work
Plan a survey about opening hours. Name the whole group, how you would select respondents, and who might still be missed.