5(j)–(k); 3i Section 5 - Evidence-based conclusions and evaluation

Drawing supported conclusions

A conclusion should answer the investigation's actual aims, questions or hypotheses rather than list every observation again. Select the findings that carry the argument, explain their significance and reach a proportionate judgement. “There was a pattern” is too vague, while “the theory is proven everywhere” extends beyond what a local investigation can normally establish.

Bring together primary and secondary evidence where relevant. Field observations may show a present pattern, while a historical record indicates whether it is recent or persistent. Interview accounts may explain an experience hidden by numerical indicators. The sources need not agree perfectly; their differences may identify distinct scales, groups or dimensions of the issue. Explain that relationship rather than selecting whichever source produces the preferred conclusion.

An illustrative access investigation finds shorter average journeys after a service moves, but some older residents report new transport difficulties. A defensible conclusion could identify improved average proximity alongside unequal practical access. The judgement depends on the evidence's definitions and coverage; it cannot simply label the change an unqualified success or failure. Different criteria may lead to different but compatible conclusions.

Use theory to interpret the result where appropriate. If a systems model predicts sediment transfer but a local structure interrupts it, the conclusion can explain the conditions under which the expected relationship holds. Evidence that qualifies a model can extend understanding more than an unsupported claim of perfect agreement. A null or unexpected result is not automatically a failed investigation if the evidence and reasoning are sound.

Keep the distinction between association and causation visible. A before-and-after difference may coincide with a management intervention but also with different weather, economic conditions or sampling. State the strongest supported explanation and the remaining alternatives. The conclusion should have a clear position, with qualifications that arise from actual evidence rather than vague uncertainty about everything.

Scope, uncertainty and wider context

A conclusion has a domain: the places, populations, periods and conditions supported by the evidence. A small local study may reveal a plausible process while providing limited evidence of its prevalence elsewhere. State that domain explicitly. An observation about one managed beach during a calm month is not automatically a conclusion about all coasts during storms.

Generalisation is stronger when the investigated mechanism and relevant conditions are understood. A finding about impermeable surfaces and runoff may be meaningful beyond one street, but slope, drainage design, rainfall intensity and maintenance can modify the effect. Compare the studied setting with the wider theory or literature before extending the claim. Avoid treating a place name as sufficient context: relevant physical and social conditions matter.

Uncertainty can affect both whether a pattern is established and how it is explained. Sampling variation may leave the population pattern uncertain; measurement error may affect its size; an omitted variable may weaken a causal explanation. A significance test addresses a specified statistical question under assumptions, not every kind of uncertainty. Failure to reject a null should not be rewritten as proof that a process has no effect.

An investigation extends geographical understanding when it shows something reasoned about the local system and its relationship to wider concepts. It may identify a neglected sediment source, reveal that environmental improvement and belonging change differently, or show why an idealised relationship depends on scale. This contribution should be explained through the evidence rather than announced as “new knowledge” merely because the learner visited the site.

Use comparison carefully. A published case may support a similar mechanism, but different methods or boundaries can limit direct numerical comparison. Agreement can strengthen plausibility; disagreement may refine the explanation. In either case, the wider context helps show what the local enquiry contributes and where its conclusions remain conditional.

The USGS 2024 Greater Mooses Tooth study compared stream samples from 2010 and 2023, including developed catchments and a reference site on Alaska's North Slope. Water-quality characteristics were generally similar across those snapshots. The study supports a conclusion about the sampled conditions; the gap between observations and limited seasonal coverage still constrain claims about all intervening events or every developed stream. A bounded conclusion preserves the finding while identifying what remains unobserved.