Practise the method.
Example 1
Two independent random samples each contain 50 students. In Sample A,26 walk to school; in Sample B,8 do. Find Sample B’s percentage minus Sample A’s percentage, in percentage points.
Show worked answer
Answer: −36
- A gives 5026×100=52%;
- B gives 508×100=16%.
- Subtract A from B to get−36 percentage points.
- Different fair samples can give different estimates without either survey being dishonest.
Example 2
Plans A and B use the same fair random method, complete population list and full response. A samples 30 students; B samples 150. Which plan’s estimates would usually vary less over repeated samples?
Show worked answer
Answer: Plan B
- The larger random sample is Plan B(150).
- With the same sound method, larger samples usually reduce sampling variation.
- This does not mean every single larger sample is closer to the truth.
Example 3
Plan A randomly samples 60 people from the complete target list. Plan B collects 9000 voluntary responses to an open web poll. All selected people respond. Which plan better addresses self-selection bias?
Show worked answer
Answer: Plan A
- Plan A selects people randomly from the complete frame.
- The voluntary poll allows people to select themselves.
- A large sample can still be systematically unrepresentative.