The Hong Kong Polytechnic University · FACULTY OF STATISTICS

RS517 Chap.3 Sampling and Survey Design

- one subject, every graph, every model, every mark
6 Chapters3-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 3 of 10 · RS517

Sampling and Survey Design

Three populations

The target population is named in the question, the sampling frame is the operational list from which selection occurs, and the achieved sample supplies data.

Probability sampling controls selection from the frame, not whether the frame covers the target or selected people respond.

Methods serve different aims

Simple random, systematic and stratified sampling use known selection routes. Cluster sampling reduces field cost but creates within-cluster similarity.

Convenience, quota and purposive methods prioritise access, chosen margins or information-rich cases and support narrower claims.

Survey error is not one number

Coverage, sampling, non-response and measurement errors arise at different stages. Increasing sample size reduces random sampling error but cannot restore missing frame members or repair a leading, double-barrelled question.

Item wording, response options and recall windows are measurement decisions.

Keep population, frame and sample separate

The target population is the group named by the claim, the sampling frame is the operational source that provides a selection route, and the achieved sample is the group that supplies usable observations. Probability selection operates only inside the frame.

It cannot restore people absent from the list or selected people who do not respond. Report coverage, invitations, responses and analysis counts as different denominators so a reader can locate every possible selection mechanism.

Choose a probability design for the field problem

Simple random sampling is transparent but can be inefficient when important subgroups are small.

Stratification can protect subgroup representation and improve precision when strata predict the outcome. Cluster sampling reduces travel or administration by selecting natural groups, but similarity within clusters reduces independent information and must enter sample-size and analysis decisions. Systematic sampling is convenient only when the ordered frame has no periodic structure aligned with the interval.

Weights must correspond to known selection probabilities, and variable weights can reduce precision.

Treat survey questions as measurements

A respondent needs one construct, a retrievable time window and response options that are complete, intelligible and ordered where order is claimed. Leading wording, hidden assumptions and double-barrelled items create measurement error that a larger sample cannot remove.

Pilot testing should examine how people interpret the item rather than only whether the form submits. Mode, language and interviewer presence can change both who answers and what an answer means.

Revision drill

Draw three nested boxes for a proposed survey and write the target, frame and achieved sample inside them. Add one route for coverage error, one for non-response and one for measurement error.

Propose a response to each and state which limitation remains. Finally, rewrite one vague satisfaction question into a behaviourally specific item with a defined recall period.

In this chapter

What this chapter covers

  • 01

    Target population, frame and achieved sample

  • 02

    Simple and systematic random sampling

  • 03

    Stratified and cluster sampling

  • 04

    Convenience, quota and purposive routes

  • 05

    Coverage and non-response error

  • 06

    Survey item construction

Worked example · free

Audit a phone-number frame

Q [8 marks]. Practice note: the marks used here are not part of the university's published assessment scheme. Only seventy per cent of discharged patients have a current number. A random sample is taken from that list. What population is represented?
The nested sets show how a random selection can still exclude target members who never appear in the sampling frame.The nested sets show how a random selection can still exclude target members who never appear in the sampling frame.Target populationSampling frameSelected samplenot covered
The nested sets show how a random selection can still exclude target members who never appear in the sampling frame.
  • 4Separate target population from frame.
  • 4Identify coverage and response checks.
Random selection represents the current-phone frame under response assumptions, not automatically all discharged patients. Compare available characteristics of covered and uncovered people, report response separately and consider a second contact route before making the broader service claim.
Sia tip — Write the target, frame and sample as three separate lines before using the word representative.
Glossary

Key terms

Target Population
The full group to which the research question intends to apply.
Sampling Frame
The operational source from which units can actually be selected.
Stratified Sampling
Probability selection performed separately inside prespecified subgroups.
Cluster Sampling
Selection of natural groups followed by observation of members within them.
Coverage Error
Distortion caused when target members have no route into the sampling frame.
FAQ

Sampling and Survey Design FAQ

Does a large sample remove sampling bias?

A large sample can reduce random sampling error around the estimate produced by its recruitment route. It does not restore target members missing from the frame, correct systematic non-response or repair measurement error. Precision can therefore increase around an estimate that still represents the wrong group.

When should stratified sampling be used?

Use stratification when important subgroups must be represented or when sampling separately within them can improve precision. Define strata before selection, sample by probability within each and use analysis weights when selection fractions differ. Stratification is not the same as post hoc subgroup analysis.

What makes a survey item interpretable?

An interpretable item asks one construct, uses a time window respondents can retrieve and offers complete, non-overlapping response options. Split double-barrelled questions, avoid wording that signals the desired answer and pilot the item with people who resemble the target respondents.

Study strategy

Exam move

Draw nested boxes for target population, sampling frame and achieved sample. For each method, add the selection unit and the principal cost or bias. Rewrite three weak survey items by splitting constructs and specifying a recall window, then explain which error each repair addresses.

Working through Sampling and Survey Design in RS517? Sia is AskSia’s AI Statistics tutor — ask any RS517 Sampling and Survey Design question and get a clear, step-by-step explanation grounded in how RS517 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 24 of your subjects at The Hong Kong Polytechnic University, plus 1,000+ more Bibles across the AskSia library.
Sia - your RS517 tutor, unlimited, worked the way the exam marks it
The full 3-page Bible + practice bank with worked solutions
Chrome extension - sync your LMS so Sia knows your deadlines
Bilingual EN / Chinese on every Bible and every Sia answer
$0.99 Trial
30-day money-back · cancel in one tap · how it works