Understand · explore · practise
Distinguish opportunity, quota and stratified random sampling. Explain practical advantages, identify selection bias and improve survey plans with original worked examples.
Before you startPopulations, sampling bias and stratified sampling.
01 / The selection process matters
Opportunity: choose convenient eligible units. Quota: fill pre-set group counts using non-random selection.
Stratified random sampling uses random selection within each group.
People can describe a survey as “random” simply because they did not know who would walk past. That is not a specified random selection mechanism. Ask who could be selected, how the choice was made and whether group targets were used.
Ask the first 20 eligible people leaving a station, with no group targets.
Choose a classification, then check.
Without the controls: the first scenario is opportunity sampling because people are chosen for availability, without group targets. The worked examples below explain all three methods.
02 / Opportunity sampling
An opportunity sample uses eligible people or items available at the place and time of collection. It is often quick and inexpensive, but the accessible group may differ from the target population. The researcher’s choices can also influence who is approached.
This is an opportunity sample
Availability at that time and place determines access.
People travelling at other times or from other stops are absent
They may have different experiences.
Useful as a quick initial investigation
It cannot automatically represent all passengers.
A student measures the first ten fallen leaves found beside the school entrance. Name the method and a limitation for estimating the mean length of all fallen leaves in the grounds.
The location restricts what is available.
Opportunity sampling. Leaves near the entrance may come from particular trees or differ from leaves elsewhere, so the sample may be unrepresentative.
03 / Quota sampling
Targets are 15 weekday-only and 10 weekend-only members
These match the known group proportions.
Interviewers approach convenient eligible members
They stop accepting a group once its quota is full.
Selection within groups is non-random
This is quota sampling, even though the allocation is proportional.
A population is 70% group A and 30% group B. Find proportional quota targets for 40 interviews.
Multiply the total by each proportion.
A: 28; B: 12. These targets alone do not specify random selection.
The quota for group A is full but a new eligible A member offers an interview. What normally happens under the stated fixed-quota plan?
The target number is already reached.
The interviewer does not add another A response to that quota sample, and continues seeking members of groups whose targets remain unfilled.
04 / Quota versus stratified
Both methods can divide the population into categories and use the same numerical allocation. Stratified random sampling selects each group’s units by a random process from suitable frames. Quota sampling lets availability or interviewer choice determine who fills each category.
Pause, replay or seek freely. The notes explain the same idea and stay in view.
Both surveys collect 18 day-shift and 12 night-shift replies. Survey A selects names randomly from each shift register. Survey B asks willing staff at the exit until its counts are full. Classify each.
Counts do not determine the method.
A is stratified random sampling, assuming the stated random selection is followed. B is quota sampling.
Why can quota interviewing often proceed without a complete list of named population members?
Targets can use known proportions without identifying every eligible person in advance.
Interviewers can classify people they encounter and fill the quotas. Reliable category information is still needed, and lack of a complete list does not remove coverage or selection bias.
05 / Representation has limits
A quota sample may match the population’s age groups but still overrepresent people who are available at lunchtime, willing to speak, or easy for interviewers to approach. People within a category can have different views. Matching category totals is not proof of representativeness.
A quota survey matches age-group proportions but interviews only at a weekday shopping centre. Give a possible remaining problem.
Think about availability within each age group.
People working elsewhere at that time are less likely to be available. Even within the same age group, their experiences or opinions may differ.
Interviewers replace every refusal with the next willing person from the same quota group. Why might bias remain?
Willingness to answer may relate to the opinion being measured.
Respondents may differ from refusers. Filling the numerical quota does not show that refusal-related bias has disappeared.
06 / Explain advantages in context
Opportunity sampling can be feasible when there is no accessible full frame and little time. Quota sampling can provide planned representation of specified categories without constructing full within-group lists. Neither advantage justifies claiming the results are automatically generalisable.
A team pilots the wording of a new questionnaire with eight available colleagues before planning its main survey. Give a useful purpose and a limit.
Testing comprehensibility differs from estimating a population percentage.
The pilot can reveal unclear wording or awkward response options quickly. The eight colleagues do not necessarily represent the main target population, so their responses should not be treated as reliable population estimates.
07 / Improve the design
Collecting at more places and times can widen access, but does not automatically turn an opportunity or quota sample into a probability sample. If a suitable frame exists, use a defined random design. Keep questions neutral and record the limitations of the actual data collected.
A town survey uses one street on one morning. Suggest one improvement to access and state what it does not guarantee.
Broaden who is available.
Use several locations and times to reach more groups. This may improve coverage, but selection can remain non-random and representativeness is not guaranteed.
Why is “just ask twice as many people at the same place and time” an incomplete repair?
The same groups can remain excluded.
It increases size but may preserve the original coverage and selection problems. A large non-random sample can still be biased.
08 / Keep the population in view
Sampling visitors at an event could be relevant to those visitors’ experience of that event. It is a weaker basis for claiming what all residents think. State the population before deciding whether the selection location is appropriate.
A cafe asks customers present at noon about today’s lunch service. Why would using the same responses to describe all local residents’ restaurant preferences require caution?
The intended population changes.
The respondents are current cafe customers at a particular time, not all local residents. People who choose other restaurants or eat elsewhere are missing.
09 / Identify targets and selection
A researcher needs 9 apprentices and 21 other staff. She interviews whichever eligible people are nearest until both totals are reached. State the method and how to convert the plan to stratified random sampling.
Keep the group counts but change the within-group selection.
It is quota sampling. To use stratified random sampling, obtain suitable current lists for the two groups and select 9 distinct apprentices and 21 distinct other staff randomly within those lists, with an explicit nonresponse plan.
Section 1 of 9 · The selection process matters