How to Define Your Population and Choose the Right Sample

Who Exactly Should Be in Your Study?

You have a research question, perhaps even a methodology and a data-collection instrument. Then comes a deceptively simple question: Who should you actually study?

This is where terms such as population, sample, target population, accessible population, and sampling frame begin to appear. They are related, but they do not describe the same thing. More importantly, choosing participants is not simply a matter of finding enough willing people. Your sample needs to make sense in relation to the question you are asking and the population about which you ultimately want to draw conclusions.

The challenge is therefore not merely deciding how many participants you need. It is establishing a defensible chain from research question → population → accessible participants or units → sampling approach → final sample.

Define the Population Before You Choose the Sample

From the Population You Care About to the Sample You Actually Study

Start With the Research Question, Not With Whoever Is Available

Sampling decisions should begin with the phenomenon and population implied by your research question. Ask: Whom or what am I trying to understand?

If you want to estimate the prevalence of academic burnout among undergraduate nursing students at a university, your population concerns undergraduate nursing students within the scope you have specified. If you want to understand how experienced school principals make decisions during organizational crises, the relevant population is defined differently. If you are analyzing research articles rather than people, the population consists of documents or other units of analysis.

This distinction matters because a research population does not necessarily mean a population of human participants. Depending on the study, the units may be households, schools, organizations, patient records, publications, social-media posts, events, or other entities.

The population should therefore follow from what the study is trying to learn. Beginning instead with an easily available group and only afterward constructing a research question around that group reverses the logic of research design.

Define the Population Precisely Enough That Membership Is Clear

A statement such as “the population consists of teachers” is usually too broad to guide sampling. Which teachers? Where? Teaching at what level? During what period? With what characteristics relevant to the research question?

A useful population definition commonly specifies the characteristics that determine whether a unit belongs to the population. Depending on the study, these might include role, setting, geographic location, institutional affiliation, experience, exposure to a phenomenon, or a relevant time period.

For example:

Too vague: College students.

More useful: Undergraduate students enrolled in fully online degree programs at the university during the second semester of the academic year under study.

The second definition makes it much easier to determine who belongs, who does not, and what sources might be used to identify potential participants. The exact boundaries should be justified by the research question rather than added merely to make recruitment easier.

Separate the Population You Want to Understand From the Population You Can Reach

Researchers often discover that the population they conceptually want to study is broader than the group they can realistically access. That difference should be recognized rather than quietly ignored.

Your intended population and the population you can actually access may differ because of geography, institutional permissions, available records, recruitment channels, cost, time, or other practical constraints.

Suppose your research question concerns public-school teachers in an entire region, but you have permission to recruit only from five schools in one division. The five schools do not become equivalent to the regional population simply because they are the sites available to you. The distinction has consequences for sampling and, later, for the scope of your conclusions.

The conceptual difference between a sample and the population it is intended to inform is fundamental. Studying a subset is often necessary, but conclusions about a larger population require a defensible basis for moving from what was observed in the sample to what is claimed about people or units beyond it.

Specify Who Is Eligible Before Recruitment Begins

Once the population is conceptually defined, translate that definition into operational eligibility rules. These rules determine who can enter the study and, when appropriate, who should not.

For example, a study of university instructors' experiences using generative AI in teaching might require participants to be currently teaching at least one higher-education course and to have used a generative AI application for a specified instructional purpose. Someone who has only read about generative AI would not necessarily belong in the same population if direct experience is central to the research question.

Your inclusion and exclusion criteria should make eligibility reproducible rather than subjective. At the same time, every restriction narrows the population represented by the study. Criteria should therefore have substantive or methodological justification rather than being added reflexively.

Determine How Members of the Population Can Actually Be Identified

Defining a population does not automatically give you a way to reach its members. You need some mechanism for identifying, approaching, or recruiting eligible units.

In survey research, this may involve a sampling frame: a list or other operational source from which potential respondents can be selected. The American Association for Public Opinion Research describes a sampling frame as the information that allows potential respondents to be contacted from a population. Depending on the study, this might be a student enrollment list, employee roster, membership registry, address database, telephone list, or panel.

The frame matters because it may not perfectly cover the population. A university email directory, for example, cannot represent eligible students who are missing from that directory. A professional-association membership list does not automatically represent every professional working in that field.

When a sampling frame does not adequately match the intended population, some eligible units may have little or no opportunity to enter the sample. This is a coverage problem, and increasing the number recruited from the same incomplete frame does not necessarily solve it.

Choose a Sampling Strategy Based on the Inference You Need

Only after clarifying the population, eligibility, and realistic source of participants should you decide how units will be selected.

The broad distinction is between probability and non-probability sampling. In probability sampling, selection is governed by a probability mechanism in which units have known selection probabilities under the design. In non-probability sampling, participants or units are selected without such known probabilities, for example through purposive recruitment, convenience access, quotas, volunteer participation, or network-based approaches.

The labels “probability” and “non-probability” are only the beginning. Specific designs include simple random, systematic, stratified, cluster, convenience, purposive, quota, snowball, and other approaches. The appropriate choice depends on the research objective, population structure, available frame, resources, analysis, and type of inference sought.

Do Not Choose Sample Size Before You Know What the Sample Is Supposed to Do

Researchers sometimes jump from “Who should I study?” directly to “How many respondents do I need?” A sample-size calculation cannot repair an ill-defined population or an unsuitable recruitment process.

For quantitative studies, the appropriate sample size and statistical power depend on the planned analysis, desired precision or power, assumptions about the expected effect or parameter, design features, and anticipated data loss or nonresponse. Different quantitative objectives therefore produce different sample-size requirements.

Qualitative research follows a different logic. Participant selection is usually tied to the phenomenon, analytic approach, heterogeneity of participants, richness of the data, and the informational needs of the study rather than a conventional statistical power calculation.

Keep the Intended Conclusions in View Throughout the Sampling Process

The population and sample are connected to what you will eventually be allowed to say. If your evidence comes from a narrow accessible population or a highly selective sample, your conclusions should reflect that boundary.

This does not mean that every worthwhile study needs a statistically representative sample. Some research seeks population estimates, while other research aims to understand mechanisms, experiences, processes, cases, or theoretical patterns. Qualitative studies, case studies, experiments, and exploratory research may pursue forms of inference that differ substantially from those of a probability survey.

The important point is alignment: the claims you make should be compatible with how the sample was constructed and what the study was designed to accomplish.

From a Broad Research Idea to a Defensible Sample

Notice that “How many instructors should I survey?” appears relatively late in this process. That is deliberate. A numerical sample-size target becomes meaningful only after the researcher knows who belongs to the population, how those people can be reached, and what the resulting data are intended to support.

Common Mistakes When Defining a Population and Sample

Build Your Sampling Plan as a Chain of Decisions

Before recruiting anyone, write down the complete path from your research question to your final sample. Each step should follow logically from the previous one. If one link is difficult to justify, that is where your design needs more attention.

A useful final test is to complete this sentence before data collection:

“We want to learn about ______, we can identify or recruit eligible units through ______, we will select them using ______ because ______, and this design allows us to make conclusions about ______ within the following limitations: ______.”

If you cannot complete that statement coherently, calculating a sample size is probably premature.

Before You Finalize Your Population and Sampling Plan

Questions About Populations and Samples in Research

Your Sample Should Follow From the Population, Not the Other Way Around

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.