Should Geography Be Part of the Target Population Definition?

Does Collecting Data in One Place Mean Your Target Population Must Be Limited to That Place?

Suppose you investigate undergraduate students' attitudes toward generative AI and collect data from universities in Metro Manila.

Should your target population be defined as undergraduate students in Metro Manila, undergraduate students in the Philippines, or university students more generally?

Now consider a different study examining the implementation of a particular local education policy. In that situation, the geographic jurisdiction may be central to the research question rather than merely the location where data happen to be collected.

Geography can therefore play several roles in research. It may define the population of interest, determine exposure to a policy or environment, influence recruitment feasibility, or simply describe where observations were obtained.

The methodological challenge is determining which role geography actually plays before incorporating it into the target population definition.

Include Geography When It Defines the Intended Population or Matters to the Research Question

How Geography Shapes Population Definitions and Research Inference

What Is a Geographic Target Population?

A geographic target population is a population whose membership is defined partly by location or a geographic jurisdiction.

For example, a study might target public secondary school teachers employed in the Philippines during a specified academic year.

Another might target households residing in a particular province or patients receiving care within a defined regional health system.

The geographic boundary may be national, regional, provincial, municipal, institutional, or based on another spatial definition.

Importantly, geography alone does not completely define the population. The researcher must also specify the relevant units, substantive eligibility characteristics, and reference period.

The broader principles of defining a research target population remain applicable.

When Is Geography Scientifically Necessary?

Geography is substantively necessary when the research question concerns a particular place, jurisdiction, environmental condition, or spatially defined population.

For example, evaluating implementation of a provincial education policy requires attention to the jurisdiction in which the policy applies.

Similarly, investigating exposure to a geographically concentrated environmental hazard may require defining the population according to residence or location during the relevant exposure period.

In these cases, geographic membership contributes directly to the scientific meaning of the study.

Removing the geographic boundary would change the population and potentially the research question.

When Is Geography Merely a Recruitment Constraint?

Geography may simply describe where the researcher can collect data.

For example, a researcher studying undergraduate attitudes toward AI may recruit only from universities near their workplace.

The accessible population is geographically restricted, but the scientific question may concern a broader student population.

Researchers should not automatically redefine the target population around the available sites merely because recruitment is geographically convenient.

Instead, they should distinguish the intended population from the population reachable through the study.

This distinction becomes particularly important when the accessible population is narrower than the target population.

How Do Geographic Boundaries Differ From Study Sites?

A geographic boundary defines the spatial scope of a population. A study site identifies a setting where data are collected or obtained.

For example, the target population might include all undergraduate students enrolled in a province, while data collection occurs at selected universities within that province.

The province defines the geographic population boundary. The participating universities are study sites.

These are related but distinct concepts.

A researcher can collect data at several sites within one geographic population or recruit participants from one site to investigate a broader question under appropriately justified assumptions.

The relationship between the population and sites should be established before making geographic generalizations.

What Geographic Dimensions Might Matter?

These dimensions should not be added mechanically. The relevant geographic characteristics depend on the phenomenon and intended inference.

Should the Population Be Defined Nationally, Regionally, or Locally?

The appropriate geographic scope follows the research question and the intended application of the findings.

A national prevalence study requires a population definition that includes the relevant national units.

A regional policy evaluation may appropriately target institutions or individuals within the policy jurisdiction.

A local case study may focus on one community because its specific circumstances are central to the investigation.

None of these scales is inherently more rigorous.

The important consideration is whether the geographic boundary supports a meaningful estimand and whether the study design can provide evidence about that population.

Can Geography Affect the Outcome Being Studied?

Yes. Geographic context may influence outcomes through infrastructure, environmental conditions, economic opportunities, institutional arrangements, and access to services.

For example, students' opportunities to use AI-supported learning systems may vary with connectivity, electricity reliability, institutional resources, and local educational policies.

However, geographic labels such as urban or rural should not be treated as explanations by themselves.

Researchers should identify the underlying mechanisms that plausibly connect location to the outcome.

Geographic differences may reflect multiple interacting social and institutional factors rather than location operating as a single causal influence.

Does Geographic Diversity Automatically Improve Representativeness?

No. Including sites from several regions may increase contextual coverage, but it does not automatically establish a probability sample of the intended population.

For example, selecting one conveniently accessible university from each region may produce a geographically distributed sample while still excluding many institutional types and student populations.

Geographic diversity and statistical representativeness are different properties.

Population inference depends on how regions, institutions, and individuals are selected, as well as coverage and nonresponse.

Researchers should avoid assuming that a map containing many study locations is sufficient evidence of national representativeness.

How Should Geographic Sampling Be Designed?

When geographic representation is important, researchers may consider stratified sampling, multistage sampling, area sampling, or other suitable designs.

For example, a national school survey might stratify schools by region and institutional type before selecting schools and students.

Alternatively, area sampling may select geographic units and subsequently identify eligible households or individuals within them.

These designs require attention to selection probabilities, coverage, clustering, and appropriate weighting.

Researchers should evaluate whether the sampling frame adequately covers the geographic target population.

A geographically broad frame may still be incomplete if certain communities or institutions are omitted.

Should Every Region Be Represented?

Not necessarily. Whether every region requires direct representation depends on the estimand and sampling design.

For example, a probability sample designed to estimate a national mean may not require an identical number of observations from every region.

However, reliable region-specific estimates may require sufficient observations within each region.

Researchers should distinguish national estimation from regional comparisons.

Sampling allocation may involve proportional selection, disproportionate stratification, or oversampling of smaller regions, followed by appropriate weighting.

The design should reflect the intended estimates rather than an arbitrary requirement for equal regional sample sizes.

What If Geographic Areas Differ Greatly in Population Size?

Geographic areas may contain very different numbers of eligible individuals.

For example, a metropolitan region may contain substantially more university students than a sparsely populated province.

Giving every geographic area equal analytical weight may estimate a different quantity from giving every individual equal weight.

Researchers should distinguish an average across geographic areas from an average across individuals.

In probability sampling, selection probabilities and weights should reflect the intended population estimand.

Equal numbers of respondents per region can be useful for regional comparisons, but they do not automatically produce a self-weighting national sample.

How Does Geographic Clustering Affect Statistical Analysis?

Observations from nearby locations may share environmental, institutional, or social influences.

For example, schools within one district may experience similar administrative policies or infrastructure conditions.

These shared influences can produce dependence among observations.

Researchers should distinguish conventional institutional clustering from spatial correlation, which may extend across administrative boundaries.

Depending on the design, survey design-based estimation, multilevel models, spatial models, or suitable clustered variance estimators may be appropriate.

Ignoring relevant geographic dependence may affect statistical uncertainty.

However, correcting spatial dependence does not automatically resolve geographic selection bias or inadequate population coverage.

Can Geographic Restrictions Reduce Confounding?

Restricting research to a geographic area may reduce variation in certain contextual characteristics.

For example, studying schools within one jurisdiction may reduce variation in formal education policy.

However, geographic restriction does not automatically eliminate confounding, because important differences may remain within the selected area.

It can also reduce the applicability of findings to settings with different conditions.

Researchers should evaluate whether geographic restriction is necessary for the question or whether relevant contextual variables can be measured and addressed analytically.

Restriction should not be used reflexively as a substitute for a defensible causal design.

What If Participants Move Between Geographic Areas?

Mobility can complicate population membership and exposure definitions.

For example, students may reside in one province while attending university in another. Patients may live outside the catchment area of the hospital where they receive care.

Researchers should specify whether geographic membership is based on residence, institutional enrollment, employment, service use, or another criterion.

In longitudinal research, individuals may change locations during follow-up.

The protocol should clarify how mobility affects eligibility, exposure classification, and interpretation.

A geographic definition that appears straightforward may become ambiguous if the relevant location and reference period are not specified.

Should Geographic Boundaries Be Defined Using Administrative Areas?

Administrative boundaries can be useful because they correspond to available records, policy jurisdictions, and recognized geographic units.

However, they are not always the most scientifically relevant boundaries.

For example, environmental exposures may extend across municipal borders, while hospital catchment areas may not correspond neatly to administrative divisions.

Researchers should select boundaries that reflect the phenomenon being studied.

Where administrative units are used for convenience, the rationale and potential limitations should be stated.

Spatial definitions should be sufficiently precise that population membership can be determined consistently.

What Is the Modifiable Areal Unit Problem?

The modifiable areal unit problem, commonly abbreviated MAUP, concerns how statistical summaries and relationships can change when geographic data are aggregated using different spatial boundaries or scales.

For example, an association observed across provinces may differ when the same underlying observations are aggregated by municipalities.

These differences can arise because geographic aggregation changes the composition and variation of the units being compared.

MAUP is especially relevant to ecological and spatial analyses using area-level data.

Researchers should not assume that findings based on one geographic partition necessarily remain unchanged under another.

Where feasible, sensitivity analyses using alternative spatial definitions may help assess the robustness of conclusions.

Can Geographic Boundaries Create Ecological Fallacies?

Yes, when researchers use area-level relationships to make unsupported claims about individuals.

For example, regions with higher average digital infrastructure may also have higher average academic achievement.

That area-level association does not establish that individual students with better internet access necessarily achieve higher scores.

Area-level relationships can differ from individual-level relationships because of aggregation, confounding, and variation within geographic units.

Researchers should distinguish geographic units of analysis from individual units and avoid extending conclusions across levels without appropriate evidence.

How Does Geography Affect External Validity?

Findings obtained in one geographic setting may not apply identically elsewhere.

For example, the effectiveness of an AI-supported instructional intervention may depend on local infrastructure, institutional support, language practices, or educational policy.

Geographic differences are especially relevant when they correspond to characteristics that modify the outcome or treatment effect.

However, geographic difference alone does not prove that an effect is different.

Researchers should identify plausible effect modifiers and evaluate whether the study population adequately represents them.

Transportability requires substantive and empirical justification rather than assuming either universal applicability or complete geographic uniqueness.

Can a Study Conducted in One City Have a Broader Target Population?

Yes. The research question may concern a broader population even when data collection occurs in one city.

However, the evidence may not support direct population-wide estimation beyond that city.

For example, a researcher might use a city-based sample to investigate a theoretical relationship relevant to university students more generally.

The study could provide useful evidence about that relationship while remaining uncertain about its magnitude across different geographic settings.

Researchers should distinguish the intended theoretical relevance from the population actually represented by the sampling design.

The location of data collection should not be confused with proof of broader generalizability.

When Should Geography Be Excluded From the Target Population Definition?

Geography need not be a defining eligibility characteristic when the research question does not require a particular location.

For example, a methodological study examining the reliability of a research instrument may not inherently concern one country or region.

Nevertheless, the locations represented in the validation sample remain relevant to interpreting measurement performance.

Similarly, a theoretical study may aim to investigate a relationship that is not defined geographically, while acknowledging that empirical evidence comes from specific settings.

Researchers should avoid introducing geographic restrictions solely because the sample happens to be collected locally.

At the same time, they should not imply that evidence from one location automatically applies everywhere.

How Should Geographic Population Boundaries Be Written?

A clear definition should specify the relevant population unit, geographic criterion, and reference period.

For example:

This definition uses institutional location rather than student residence as the geographic membership criterion.

A different research question might instead define the population according to students' usual residence.

The important requirement is to specify which geographic relationship determines eligibility and why that relationship is relevant.

How Should Geography Be Reported in Research?

Methods sections should identify the geographic scope of the target population, study sites, and recruitment arrangements.

Describe the geographic membership criterion, relevant reference period, and any restrictions in coverage.

For probability-based studies, report geographic sampling procedures, stratification, selection probabilities, and weighting where relevant.

For purposive or convenience-based studies, explain why particular locations were selected and how they relate to the research question.

Where spatial dependence or aggregation matters, describe the geographic units used in analysis and any relevant sensitivity checks.

Clear reporting allows readers to distinguish geographic population boundaries from practical limitations of data collection.

Defining Geography in a Study of University Students' AI Literacy

Suppose 80% of participating respondents achieve a predefined AI literacy threshold.

That observed proportion does not automatically represent the percentage of undergraduate students nationally who meet the same threshold.

The six universities may differ from other institutions in ways associated with AI literacy, and their convenience selection does not provide known national institutional inclusion probabilities.

However, the researcher could legitimately revise the study to focus on AI literacy within the six participating universities, provided the research question and population claims are adjusted transparently.

Alternatively, a broader national target population could remain the substantive motivation while the study is explicitly presented as geographically restricted evidence.

The important distinction is between geography as part of the scientific population definition and geography as a limitation of the recruitment process.

Common Misconceptions About Geographic Population Boundaries

How to Decide Whether Geography Belongs in Your Population Definition

Begin by asking whether location is essential to the phenomenon, eligibility, policy context, or intended population claim.

Then distinguish geographic boundaries from the practical locations available for recruitment.

Geographic boundaries should clarify the research question, not merely describe where the researcher happened to obtain permission to collect data.

What to Verify Before Defining a Geographic Target Population

Frequently Asked Questions About Geography and Target Populations

Geography Should Define the Population When Location Matters to the Question

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.