EDS6003 Chap.5 Sampling and Representativeness
Sampling and Representativeness
Selection governs the population claim
The sampling lecture defines a population as the complete specified group and a sample as a subset. Representativeness concerns whether values and patterns found in the sample also exist in that population. Sample size improves precision but does not repair systematic exclusion.
The selection process and response pattern therefore matter before any statistical calculation.
Different routes support different inferences
Simple random, systematic, stratified and cluster sampling use probability-based procedures. Convenience, snowball and purposeful selection serve access or information-rich case selection when probabilities cannot be calculated.
They may be appropriate for qualitative depth or hidden groups, but they do not license population percentages. Good reporting distinguishes the sampling frame, invited sample, nonresponse and achieved sample.
What this chapter covers
- 01
Population and sample
- 02
Representativeness
- 03
Simple random sampling
- 04
Systematic, stratified and cluster sampling
- 05
Convenience and snowball sampling
- 06
Purposeful selection and bounded claims
Worked example · free
Protect subgroup representation
- 3Choose and justify a probability design.
- 3Describe selection within groups.
- 3Explain how response may alter the sample.
Key terms
- Population
- The complete specified group about which the study intends to draw a conclusion.
- Sample
- The subset of a population from which observations are actually collected.
- Sampling Frame
- The list or operational route that gives access to potential sample members.
- Stratified Sampling
- Probability selection performed separately within non-overlapping subgroups to protect their representation.
- Purposeful Sampling
- Non-probability selection of information-rich cases using substantive knowledge of the population.
Sampling and Representativeness FAQ
Does a large sample guarantee representativeness?
No. A large sample can estimate its own pattern precisely while systematically excluding important groups. Representativeness depends on the relation between population, frame, selection and response. Size cannot turn an open convenience link into a probability sample.
When is purposeful sampling appropriate?
Purposeful sampling is appropriate when cases are selected because they possess experience or variation needed to illuminate a qualitative question. The report should state the criteria and rationale, then limit conclusions to mechanisms or meanings rather than estimating how common a view is.
What can go wrong with systematic sampling?
Selecting every kth entry can reproduce periodic structure in an ordered list, such as always choosing the same position within tutorial groups. A random start does not solve a matching cycle. Shuffle the frame, change the interval or stratify before selecting.
What should a sampling justification actually explain?
Name the target population, accessible frame, selection procedure, invitation route and achieved participants, then connect that chain to the intended claim. Probability sampling requires an operational random mechanism, while purposeful selection requires an information-rich criterion and analytic contrast. Discuss who could not enter the frame and who did not respond.
A design label alone cannot show what population or experience the evidence represents.
Assessment move
Write five lines for every proposed sample: target population, available frame, inclusion criteria, selection procedure and intended claim. Then add a sixth line for nonresponse. If the intended claim is broader than the people the frame and response can represent, narrow the claim or redesign recruitment before collecting data.
Draw the sampling chain as population, frame, selected cases, invited cases and achieved participants. Put a number beside each stage when one is knowable and write the reason for every loss. This makes coverage and nonresponse visible as different problems. For a probability design, rehearse the actual randomisation procedure with a tiny fictional list, including the random start or stratum allocation.
A design name without an operational selection rule is not reproducible. For purposeful recruitment, write the information criterion for every case and identify the contrast that the case contributes; do not substitute convenience for an analytic rationale. Stress-test the frame by listing people who meet the population definition but cannot appear on it. Predict how their absence might shift the observed pattern.
Compare response across relevant subgroups rather than treating one overall percentage as proof of quality. If a subgroup is deliberately oversampled, distinguish the composition needed for comparison from the weights needed for a population summary. End the exercise with two conclusions: the strongest statement the achieved sample can support and a tempting broader statement it cannot support.
The contrast trains the boundary that must appear in the report’s discussion and limitations.
Working through Sampling and Representativeness in EDS6003? Sia is AskSia’s AI Education tutor — ask any EDS6003 Sampling and Representativeness question and get a clear, step-by-step explanation grounded in how EDS6003 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
