FBTE64441 Chap.4 Sampling, Reliability and Validity
Sampling, Reliability and Validity
Sampling, Reliability and Validity examines two links in the inference chain: who or what is observed, and whether the intended construct is measured. Sampling moves from a target population to a frame, selection method and achieved sample. Coverage, selection and nonresponse can each narrow the supported population claim.
Reliability concerns stability or consistency, while validity concerns whether the procedure and interpretation represent the intended construct and purpose. Reliability is necessary but not sufficient because a measure can repeat the same error consistently.
The chapter combines reproducible selection calculations with an audit of the frame, then tests measurement claims through content, criterion and construct validity rather than treating one coefficient as universal proof.
What this chapter covers
- 01
Target population and sampling frame
- 02
Probability and non-probability selection
- 03
Simple, systematic and stratified sampling
- 04
Coverage, selection and nonresponse
- 05
Test-retest and internal consistency
- 06
Content, criterion and construct validity
Worked example · free
Construct a systematic sample
- 2Divide 1,200 by 100 to obtain a systematic interval of 12.
- 2Choose a random start from 1 to 12 and then select every twelfth case.
- 2Inspect the list for periodic ordering and keep the claim bounded to firms covered by the frame.
Key terms
- Sampling frame
- The operational list or source from which eligible cases can be selected.
- Reliability
- The consistency or stability of a measurement procedure or resulting score.
- Construct validity
- Evidence that a measure behaves as the intended theoretical construct should behave.
Sampling, Reliability and Validity FAQ
Why is a sampling frame important?
The frame determines which eligible cases can actually be selected. If relevant groups are absent, random selection within the list does not repair the coverage difference. Describe both the target population and frame before claiming representation.
Does a high response rate prove representativeness?
No. A response rate describes participation from the invitations used, but the original frame may still be narrow and respondents may differ from nonrespondents on relevant characteristics. Inspect coverage, recruitment and achieved-sample balance together.
Can a reliable measure still be invalid?
Yes. Repeated or highly similar items can produce stable, internally consistent scores while measuring only a narrow proxy. Validity requires evidence that the items and interpretation represent the intended construct and purpose.
How do probability and purposive samples differ?
Probability sampling gives eligible frame members a known selection mechanism and can support design-based population inference when its assumptions hold. Purposive sampling deliberately recruits information-rich cases for analytical depth or variation. Its strength is relevance to the question, not statistical representativeness.
What does nonresponse change?
Nonresponse reduces the achieved sample and may alter its composition when participation relates to the outcome or key characteristics. Report the denominator, compare respondents with available frame information and narrow the claim if imbalance is plausible. A larger invitation list does not remove selective participation.
Is reliability a property of an instrument forever?
No. Reliability evidence concerns scores produced in a particular population, setting and administration. An instrument that behaves consistently in one context may function differently after translation, format changes or use with another group. Report the relevant evidence rather than treating a coefficient as permanent certification.
How should I justify sample size?
Link the justification to the design and intended claim. Quantitative work may use precision, power or modelling requirements, while qualitative work considers information richness, variation and analytical depth. Convenience alone is not a rationale, and a large biased sample does not guarantee sound inference.
What is a validity threat?
A validity threat is a plausible process that supports an alternative interpretation of the observed evidence. Confounding can challenge causal claims, frame undercoverage can challenge population claims, and weak construct measurement can challenge meaning. Name the threatened claim and describe how design or analysis addresses it.
Exam move
Draw the sampling pipeline from target population to frame, selection method, achieved sample and final claim. For every practice scenario, identify the first link that excludes or distorts relevant cases. Compare simple random, systematic, stratified, purposive and convenience methods by their selection logic and supported inference rather than memorising labels.
Recompute systematic intervals and response rates, then state what the numbers cannot repair. Study reliability and validity as separate questions. Build examples of an unstable measure, a reliable but narrow proxy and a well-supported interpretation.
Match each form of evidence to its purpose: test-retest for stability, internal consistency for item coherence, content validity for construct coverage, criterion validity for performance against an external criterion and construct validity for theoretically expected relationships. Combine sampling and measurement in revision cases because one can be strong while the other is weak.
In exam answers, state the mechanism: frame undercoverage, self-selection, nonresponse, repeated wording or missing construct dimensions. Finish by rewriting the population or construct claim at the strength the design supports.
Working through Sampling, Reliability and Validity in FBTE64441? Sia is AskSia’s AI Management tutor — ask any FBTE64441 Sampling, Reliability and Validity question and get a clear, step-by-step explanation grounded in how FBTE64441 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.