03 · What You Need to Know
How to Define a Research Target Population With AI Assistance
What Is a Target Population in Research?
The target population is the defined collection of individuals, groups, institutions, events, or other units about which the researcher intends to draw conclusions.
In survey research, the target population establishes the population to which estimates are intended to refer. In other forms of research, the relevant population or universe of cases may be defined according to the study's theoretical and methodological purpose.
For example, a study examining generative AI use among undergraduate students at a particular university may define its target population as all undergraduate students enrolled at that institution during a specified academic period.
The definition identifies the intended population independently of whether every member can actually be contacted.
Lohr (2022) emphasizes the importance of defining populations and sampling frames when designing probability samples and interpreting survey estimates.
What Is the Difference Between Target Population, Accessible Population, Sampling Frame, and Sample?
These terms describe related but distinct aspects of the research process.
| Term |
Meaning |
Illustrative example |
| Target population |
The population about which the researcher intends to make inferences. |
All undergraduate students enrolled at University A during the academic year. |
| Accessible population |
The portion of the relevant population that the researcher can realistically reach or recruit under the study's conditions. |
Students enrolled in colleges that have authorized recruitment. |
| Sampling frame |
The operational list or other representation from which units can be selected. |
An authorized student enrollment roster. |
| Sample |
The units selected or recruited for the investigation. |
The students selected from the roster and participating in the study. |
The accessible population and sampling frame may not coincide. A researcher might be permitted to contact students through a university mailing system without receiving an individual-level roster.
Likewise, a sampling frame may omit eligible individuals or include outdated records.
AI should not treat these categories as interchangeable merely because the same institution is mentioned in each description.
Why Should the Research Question Determine the Target Population?
Population definition should follow what the study intends to establish.
Consider three questions about generative AI in higher education.
| Research question |
Possible target population |
Reason |
| What proportion of undergraduate students use generative AI? |
All eligible undergraduate students, including nonusers. |
Excluding nonusers would distort the intended prevalence estimate. |
| How do experienced AI users evaluate AI-generated feedback? |
Students meeting a justified definition of relevant AI feedback experience. |
Direct experience is central to the inquiry. |
| How do university instructors implement institutional AI policies? |
Instructors within the specified institutional and policy context. |
The research concerns instructors' professional practices rather than students' use. |
The topic may remain generative AI, but the relevant population changes with the question.
AI may identify these differences when given a precise research objective. Without that information, it may simply reproduce a broad population commonly associated with the topic.
What Boundaries Should a Target Population Definition Include?
A useful population definition identifies the characteristics necessary to determine whether a unit belongs to the population.
Depending on the study, these may include educational level, enrollment status, institutional setting, geographic scope, academic period, professional role, or relevant exposure characteristics.
For example, "university students in the Philippines" may be too broad for a study intended to investigate currently enrolled undergraduate students at public universities during a particular academic year.
However, additional restrictions should not be introduced merely to make the definition appear more precise.
Every boundary should serve the research question, population inference, or legitimate implementation requirements.
Why Does the Time Period Matter?
Population membership can change over time.
Students enroll, graduate, transfer, or withdraw. Institutions may introduce new AI policies, and access to particular technologies may change.
A study estimating AI use among students enrolled during the first semester of an academic year may therefore concern a different population from one covering all students enrolled at any point during that year.
AI should help researchers specify the relevant reference period rather than assume that the population remains unchanged.
Temporal boundaries also matter when researchers compare repeated cross-sectional samples or follow a defined cohort over time.
Can AI Make a Target Population Too Broad?
Yes. AI may suggest a population that exceeds the study's intended scope or evidentiary capacity.
For example, a researcher studying students at one private university may receive an AI-generated population statement referring to all Filipino university students.
The broader population may be substantively interesting, but data from one institution do not automatically support estimates for all universities in the country.
Differences in institutional resources, student characteristics, educational programs, and technology access may affect the applicability of findings.
The researcher should distinguish the population actually defined for sampling from any broader population to which findings might later be transported or generalized under additional assumptions.
Can AI Make a Target Population Too Narrow?
Yes. AI may recommend restricting the population to participants who are easiest to recruit or most familiar with the phenomenon.
Suppose the objective is to estimate the prevalence of generative AI use among university students.
If AI defines the population as students who have previously used ChatGPT, the resulting study cannot directly estimate the proportion of all students who use generative AI because nonusers have been excluded by definition.
Similarly, restricting participation to students with reliable internet access may exclude an important group when the research concerns unequal technology adoption.
Population restrictions should therefore be evaluated for their consequences, not merely their convenience.
Is the Target Population Always the Same as the Population Available for Recruitment?
No. A researcher may intend to study a broader population than can be reached through available recruitment arrangements.
For example, a researcher may wish to investigate all public university students in a region but have permission to recruit only from two institutions.
The accessible population is narrower than the intended target population.
That difference should be made explicit. It may require changes to the sampling strategy, additional recruitment arrangements, or more cautious interpretation.
Researchers should not silently redefine the target population after discovering that recruitment is difficult.
Instead, they should determine whether the scientific objective can be retained or whether the population and corresponding research question require justified revision.
Can AI Determine Whether a Sampling Frame Covers the Target Population?
AI can help compare the stated population definition with the information contained in a proposed sampling frame.
For example, an enrollment roster may exclude students from certain campuses or programs. If those students belong to the target population, the frame has a coverage problem.
Conversely, a roster may contain graduates or students who are no longer enrolled.
AI can identify these possible discrepancies when the relevant information is supplied, but it cannot verify the completeness of institutional records without authoritative evidence.
Coverage problems are important because a probability selection procedure applied to an incomplete frame does not automatically provide valid population coverage.
Does Defining the Target Population Establish Representativeness?
No. Population definition establishes the intended scope of inference. Representativeness depends on the subsequent sampling, recruitment, response, measurement, and analytical processes.
A clearly defined population may still be investigated through a convenience sample that systematically excludes some members.
Likewise, a probability sample may experience nonresponse that affects estimates.
The American Association for Public Opinion Research emphasizes transparent reporting of population definitions, sampling procedures, recruitment, and survey limitations.
Researchers should therefore distinguish defining the population from selecting an appropriate sampling method.
How Does Target Population Definition Differ in Qualitative Research?
Qualitative studies may define the relevant universe of participants or cases according to experiences, practices, settings, or processes central to the inquiry.
For example, a qualitative investigation of instructors' experiences implementing AI assessment policies may seek participants who have directly encountered the relevant policy.
The objective may be to obtain information-rich accounts rather than estimate the proportion of all instructors holding a particular view.
However, the population or case boundaries still require justification.
Researchers should clarify whose experiences are relevant, what contextual conditions matter, and which perspectives the sampling strategy may not capture.
Qualitative population definition should follow the methodological tradition rather than adopt statistical representativeness as a universal standard.
How Are Target Population and Eligibility Criteria Related?
The target population establishes the intended scope of the study. Eligibility criteria translate relevant boundaries into operational rules for determining who can participate.
For example, a target population may consist of currently enrolled undergraduate students at a specified university.
Inclusion criteria may then specify enrollment status, academic period, and other necessary characteristics.
Exclusion criteria may address conditions that make participation inappropriate or prevent the intended measurements, provided these restrictions are justified.
Eligibility criteria should not unnecessarily narrow the population or contradict the research question.
Researchers should separately examine how AI can help develop inclusion and exclusion criteria after the target population has been defined.
Can AI Determine the Population Size?
AI may help identify where population counts could be obtained or explain how counts should be interpreted.
However, it should not invent the number of eligible participants or assume that total institutional enrollment equals the size of the target population.
For example, a university may report 15,000 students, but the target population may include only currently enrolled undergraduate students from selected programs.
The appropriate count should be obtained from authorized and current institutional records or another defensible source.
Population size may influence sampling and precision calculations, but a numerical count does not replace a clear population definition.
How Can You Audit an AI-Generated Target Population Definition?
1. State the research question. Identify exactly which population the intended findings concern.
2. Define the unit of interest. Specify whether the study concerns individuals, institutions, classrooms, documents, or other cases.
3. Establish substantive boundaries. Identify relevant characteristics such as educational level, institutional setting, and exposure or experience.
4. Specify geographic and temporal scope. Clarify where and when population membership is defined.
5. Examine exclusions implied by the definition. Determine whether important eligible groups have been unintentionally omitted.
6. Compare target and accessible populations. Identify recruitment limitations and differences in coverage.
7. Verify population information. Confirm institutional counts, enrollment conditions, and sampling frame availability through appropriate sources.
8. Document the intended inference. Explain which population the findings will describe and which broader claims would require additional justification.
The resulting population statement should be sufficiently precise to guide sampling without introducing restrictions that the research question does not require.
Watch Out
Do not define the target population solely according to who is easiest to recruit. If the accessible participants differ from the population the study intends to describe, the discrepancy must be addressed rather than hidden through an overly broad population statement.
