03 · What You Need to Know
The Scientific Case for Local Research Is About Applicability, Not Geography
Existing evidence and applicable evidence are not the same thing
A study can provide strong evidence for the population and conditions it examined while providing less certain evidence for a substantially different target setting. This is an issue of external validity, generalizability, applicability, or transferability. These terms have somewhat different meanings across methodological traditions, but they share a central concern: how confidently can an inference supported in one set of circumstances be extended to another?
The Agency for Healthcare Research and Quality, for example, describes applicability in terms of whether effects observed in published studies are likely to reflect expected results when an intervention is applied to the population of interest under real-world conditions. GRADE similarly treats important mismatches between available evidence and the question of interest as potential sources of indirectness.
This does not mean that every contextual difference makes evidence inapplicable. GRADE explicitly cautions against treating differences as important without compelling reasons to expect meaningful changes in relative or absolute effects. The burden of the argument is therefore not to demonstrate that two settings are different. Almost any two settings are. The task is to explain why the particular difference could matter to the inference.
Geographical novelty
The question has not previously been studied in a particular country, city, institution, or community.
Contextual uncertainty
There is a plausible reason why evidence generated elsewhere may produce a different result, interpretation, implementation outcome, or decision in the target setting.
The second provides the stronger scientific rationale. Simply saying that something has never been studied in your country establishes a location gap, but not necessarily a consequential knowledge gap.
Start by defining exactly what you want to transfer
Before deciding that local evidence is needed, identify the claim you want to apply locally. Researchers sometimes discuss an entire study as though it either “applies” or “does not apply,” but transferability can concern different claims.
Perhaps international research suggests that an educational intervention improves achievement. You may have little reason to doubt the underlying direction of the effect, yet considerable uncertainty about whether schools in your setting have enough teachers, devices, instructional time, or technical support to implement the intervention as studied. In that case, the unresolved local question may concern implementation rather than whether the intervention can ever work.
Likewise, an established relationship may generalize while its magnitude differs. A risk factor could operate in the same direction across countries but have different absolute consequences because baseline prevalence differs. The need for local research therefore depends partly on what you need to know: existence of an effect, effect size, prevalence, feasibility, acceptability, cost, implementation, or likely local impact.
Population differences matter when they could modify the finding
International studies may involve populations that differ from the local target population in age, socioeconomic circumstances, prior exposure, language, educational background, disease burden, baseline risk, demographic composition, or other characteristics relevant to the research question.
Not every difference is consequential. Researchers should ask whether there is a credible mechanism by which the characteristic could affect the outcome or interact with the exposure or intervention. A demographic difference that is merely descriptive is a weaker justification than one supported by theory, prior evidence, or a plausible causal mechanism.
This is why the more precise question is whether a contextual difference is scientifically meaningful, rather than whether the local sample looks different from samples reported abroad.
Culture may affect constructs, behavior, and intervention responses
Cultural context can sometimes influence how people understand survey items, express attitudes, respond to incentives, interact with institutions, or engage with interventions. Constructs developed in one linguistic or cultural environment may also require more than literal translation before they can be interpreted equivalently elsewhere.
Yet “our culture is different” is too broad to carry the argument by itself. A defensible rationale should specify the cultural characteristic, explain the mechanism through which it could affect the variable or relationship under investigation, and identify what uncertainty follows from it. The question of when cultural differences justify new research is therefore empirical and theoretical, not merely geographical.
Institutions can change how the same intervention operates
Many findings depend partly on systems surrounding the people being studied. Healthcare delivery, curricula, teacher preparation, regulation, social protection, organizational structures, assessment systems, labor markets, and public policy can alter exposures, incentives, implementation, and outcomes.
Imagine an intervention whose effectiveness depends on referral pathways between primary and specialist care. Evidence from countries with integrated referral systems may not straightforwardly predict performance in a fragmented system. Similarly, an educational technology intervention evaluated where students have individual devices and reliable connectivity may encounter a substantially different implementation environment where devices are shared or internet access is intermittent.
These are not merely national differences. They are differences in mechanisms and conditions. That is what can make healthcare, educational, policy, or institutional context scientifically relevant.
Resource constraints can alter feasibility and effectiveness
An intervention shown to work under well-resourced conditions may not produce the same results when personnel, infrastructure, time, funding, equipment, or expertise are constrained. Resource differences can affect whether an intervention is delivered at the intended intensity or fidelity and whether it can be sustained after a study ends.
Again, resource scarcity does not automatically require another efficacy study. Sometimes the unresolved question is implementation feasibility, adaptation, cost-effectiveness, or scalability. A researcher should therefore identify which resource difference could plausibly alter the relevant outcome before deciding what type of local study is needed.
Local baseline data may be needed even when an underlying relationship transfers
International evidence can be applicable in one respect and insufficient in another. Suppose robust evidence establishes that an intervention reduces the relative risk of an outcome. Local decision-makers may still need local estimates of baseline risk to determine the likely absolute benefit.
This distinction matters because the solution is not necessarily to repeat the entire international study. GRADE notes that uncertainty about baseline risk can affect confidence in absolute effects and that supplementary evidence may sometimes resolve the problem. Local surveillance, administrative records, registries, or a smaller observational study could therefore answer the unresolved question more efficiently than a full replication.
Watch Out
Do not jump from “international evidence may not transfer perfectly” to “we need to repeat the entire study locally.” First identify exactly which inference is uncertain. The most appropriate local research design should target that uncertainty.
Decision relevance can make local evidence particularly valuable
Evidence is often collected to support a decision rather than merely to accumulate publications. A ministry may need to decide whether to fund an intervention. A university may be considering a new assessment policy. A hospital may need to know whether a service model can operate with its staffing structure.
In these situations, even a large international literature may leave locally consequential parameters uncertain. The scientific strength of the international evidence still matters, but so does its directness to the decision. Under some circumstances, local evidence may be more useful for a particular local decision than stronger but substantially less applicable evidence from elsewhere.
That does not mean weak local research should automatically outrank rigorous international research. Local relevance cannot repair serious bias. Ideally, researchers integrate what is already known internationally with the local information needed to judge applicability.
Sometimes the appropriate response is local replication
If the central scientific uncertainty concerns whether a previous finding persists under meaningfully different conditions, a local replication can be informative. Replication is not scientifically useful because the postal code changed. It becomes useful when the new conditions provide an informative test of the scope of the earlier claim.
The National Academies describes generalizability as the extent to which study results apply in other contexts or populations. Consistent findings across deliberately different contexts can therefore strengthen evidence that a phenomenon is not confined to the circumstances of the original study.
Seen this way, a carefully chosen local replication can contribute evidence about generalizability. Conversely, repeating an established finding in a nearly interchangeable setting, without a meaningful contextual hypothesis, risks reproducing knowledge that is already sufficiently secure.
Local research can also address questions the international literature has neglected
There is another reason to be cautious about allowing international visibility to determine local research priorities. The questions most heavily represented in the global literature are not necessarily the questions of greatest consequence everywhere.
A locally important disease, educational problem, environmental exposure, infrastructure constraint, or policy issue may receive relatively little international attention. In such cases, prioritizing a consequential local problem can be more defensible than reproducing a fashionable global research question whose practical relevance to the target population is limited.
The direction also works the other way. A topic may be globally prominent precisely because it matters in many settings. Researchers should not dismiss international priorities merely for being international. The point is to establish relevance rather than inherit it.
A useful test is whether the local study could change what we believe or do
One practical way to evaluate necessity is to imagine the possible results before collecting data.
If the local study reproduced the international finding, what would you learn? If it produced a different finding, would that difference be scientifically interpretable? Could you identify a contextual mechanism that might explain it? Would either result change theory, practice, policy, implementation, or confidence in generalizability?
If neither possible outcome would materially change understanding or action, the scientific value of the study may be modest. If either outcome would resolve an important uncertainty, then the case for local research becomes much stronger.
