PES6251 Chap.8 Sampling and Group Allocation
Sampling and Group Allocation
Follow every transition into the analytic sample
The target population defines who the question concerns, the accessible population defines who can be reached, and the sampling frame determines who can be selected. Probability sampling supports population inference when coverage and response are adequate. Purposive, convenience and network methods serve other aims but narrow reach.
Random selection governs representation, while random assignment governs comparability between intervention groups.
Coverage begins before invitation
A membership list, school register or digital community contains only people visible through that system. If informal coaches, absent pupils or less connected participants are missing, random selection within the frame cannot restore them.
Compare the frame with the target population and state which groups are excluded. The consequence depends on the outcome: a technology-platform frame may inflate estimated adoption because comfort with digital tools affects both visibility and response.
Response and retention create new samples
Track eligible, contacted, consenting and measured participants separately.
Non-response threatens inference when willingness relates to the variable under study, and attrition becomes especially serious when it differs by exposure or condition. Compare achieved numbers and characteristics across key groups rather than citing invitation targets.
A stratum can be properly sampled yet too small after non-response for a precise comparison, which should change the analysis and strength of the conclusion.
Match the sampling unit to analysis
Cluster sampling recruits natural groups such as clubs or schools and reduces logistical cost, but members share environments and are statistically correlated. Multistage designs should describe every selection level.
If clubs receive an intervention, the allocation unit is the club even when athlete outcomes are recorded. Planning and analysis must allow for that structure or uncertainty will be understated. A large participant count does not create the information of the same number of independent units.
What this chapter covers
- 01
Target and accessible populations
- 02
Sampling frames and coverage
- 03
Probability sampling methods
- 04
Purposive and network recruitment
- 05
Non-response and attrition
- 06
Random selection versus allocation
Worked example · free
Repair a convenience-sample claim
- 3Identify coverage and voluntary-response limits.
- 2Rewrite the target population.
- 3Propose a stronger frame and sampling route.
Key terms
- Target Population
- The full group to which the research question and intended conclusion refer.
- Sampling Frame
- The operational list or procedure from which eligible units can be selected.
- Coverage Error
- Mismatch between the target population and people who can appear in the sampling frame.
- Stratified Sampling
- Sampling separately within defined subgroups to protect representation or precision.
- Cluster Sampling
- Selection of naturally occurring groups such as clubs or schools rather than independent individuals.
Sampling and Group Allocation FAQ
Does random sampling guarantee a representative final sample?
No. Random selection protects the invitation stage, but frame gaps, failed contact, refusal and attrition can still make the observed sample selective. Report how many people were eligible, invited, enrolled and measured, then assess whether each transition relates to exposure or outcome.
Why are athletes within one club not independent observations?
They share coaches, facilities, schedules and local practices, so their outcomes tend to be correlated. Treating them as fully independent understates uncertainty. Sample-size planning and analysis should recognise the club cluster, especially when clubs rather than athletes receive an intervention.
When is purposive sampling defensible?
It is defensible when the aim is to obtain information-rich accounts from people with relevant experience rather than estimate population prevalence. State the selection criteria and sought variation. The conclusion should explain the experiences studied without implying numerical representation of everyone in the wider population.
Assessment move
Draw four boxes labelled target, frame, enrolled sample and analytic set. Add counts and reasons for loss at every arrow, then write how each transition might move the outcome. Keep a separate line for assignment after enrolment so population representativeness and internal comparison are never confused.
For each sampling method, name the list or process that gives people a chance of selection and identify who has no route into it. Distinguish a large convenience sample from a probability sample rather than using sample size as a substitute for coverage. If groups are assigned, record the unit of allocation, possible contamination and whether analysis needs to account for clusters.
Finish with one internal-validity claim and one population-reach claim.
Working through Sampling and Group Allocation in PES6251? Sia is AskSia’s AI Statistics tutor — ask any PES6251 Sampling and Group Allocation question and get a clear, step-by-step explanation grounded in how PES6251 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
