BENV0139 Chap.8 Statistical Evidence for Heritage Decisions
Statistical Evidence for Heritage Decisions
Statistical Evidence for Heritage Decisions integrates effect estimates, intervals, study quality and practical stakes. Two studies can fall on opposite sides of a significance threshold while their confidence intervals overlap and their estimates remain compatible. Binary labels can therefore exaggerate disagreement.
Sample size affects precision, but measurement noise, population heterogeneity and implementation differences also matter. A heritage decision adds consequences: low-cost reversible monitoring can be reasonable under wider uncertainty, while invasive treatment may require stronger evidence.
The published marking criteria reward precise statistical explanations, critical assumptions and limitations, realistic heritage applications and clear structure. This chapter provides a narrative architecture that begins with the decision, compares magnitude and uncertainty, investigates design differences, and ends with a conditional action plus the evidence that would change it.
What this chapter covers
- 01
Effect size
- 02
Confidence interval overlap
- 03
Study comparability
- 04
Statistical power
- 05
Practical significance
- 06
Decision costs and reversibility
- 07
Critical narrative structure and conditional recommendations
Worked example · free
Reconcile two preventive-conservation studies
- 2Compare estimates and identify interval overlap from 0.4 to 3.2. The marks are original practice weighting.
- 2Explain that Study B permits harm, no effect and benefit rather than proving absence.
- 2Investigate population, measurement and implementation differences before pooling.
- 2Tie the recommendation to cost, reversibility and a monitored evidence plan.
Key terms
- Effect size
- The magnitude of a difference or relationship expressed on a scientifically meaningful scale.
- Practical significance
- The degree to which an effect matters for a real decision, regardless of its statistical detectability.
- Decision boundary
- A stated condition under which the recommended action would change.
Statistical Evidence for Heritage Decisions FAQ
Can significant and non-significant studies agree?
Yes. Their estimates and confidence intervals may overlap substantially even when p-values fall on different sides of 0.05. Compare magnitudes, uncertainty, populations and methods before describing conflict. A binary label discards much of the evidence.
How does practical importance enter?
Define the outcome change large enough to matter for objects, buildings or stakeholders. Compare the interval with that threshold and consider intervention cost, reversibility and harm. A detectable tiny effect can be irrelevant, while an uncertain large benefit may justify a monitored pilot.
What structure suits a long exam answer?
Open with the decision and central comparison. Interpret estimates and intervals, discuss design and assumptions, assess practical consequences, then give a conditional recommendation. Headings and purposeful paragraphs help the marker see both statistical thinking and heritage application.
Exam move
Practise evidence synthesis with pairs of estimates and intervals rather than isolated p-values. For each pair, mark the overlap, the values compatible with benefit or harm, and the design differences that could explain variation. Add a heritage decision table with cost, reversibility, material risk and stakeholder priorities. Then write a recommendation that remains sensible across the plausible effect range.
Use the published criteria to self-mark: Is the statistical concept precise? Are assumptions and limitations critical rather than generic? Does the example contain real data, variables and sector consequences? Is the writing structured and scientific? Repair vague caveats by naming the direction of bias and the conclusion affected. Repair generic examples by adding material, site, outcome and decision-maker.
Under timed conditions, draft a five-line plan before prose: define, apply, challenge, interpret and decide. Finish with a decision boundary and the next observation or study that would change the recommendation. Build an interval line for every study and mark zero, the smallest worthwhile benefit and the largest tolerable harm.
Compare populations, outcome scales, follow-up, implementation and measurement quality before deciding whether estimates address the same quantity. If standardising effects, retain an original-unit translation so the recommendation remains meaningful. Create three action scenarios using identical evidence but different reversibility and downside: monitored pilot, costly system change and invasive treatment.
Explain why the evidential threshold changes. Practise seven-part answers that define the method, show the evidence route, interpret magnitude, inspect assumptions, make criticism directional, apply heritage context and state a conditional action. Replace generic phrases such as small sample or possible bias with the effect on interval width, estimate direction or transfer.
Use calibrated verbs and remove proves, safe and no effect unless the design truly supports them. End each timed response with what should happen now, the scope of the recommendation and the exact observation that would trigger escalation, continuation or stopping.
Working through Statistical Evidence for Heritage Decisions in BENV0139? Sia is AskSia’s AI Statistics tutor — ask any BENV0139 Statistical Evidence for Heritage Decisions question and get a clear, step-by-step explanation grounded in how BENV0139 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.