What Is the Unit of Analysis in a Case Study?

Is the Unit of Analysis the Person, Institution, Program, or Process Being Studied?

You are investigating how a university implemented a generative AI policy. You interview fifteen faculty members, five administrators, and ten students. You also examine policy documents, meeting records, and institutional reports.

What is the unit of analysis? Is it each participant, the university, the policy, or the implementation process?

The answer matters because the unit of analysis determines what the study is fundamentally examining and what conclusions its evidence can support.

Researchers sometimes confuse the people supplying information with the case being investigated. Others identify the institution as the case even when the actual research question concerns a specific program or decision-making process within it.

Without a clear unit of analysis, the research questions, case boundaries, sampling strategy, analytical procedures, and conclusions may become inconsistent.

The Unit of Analysis Is What the Case Study Seeks to Understand and Explain

How the Unit of Analysis Shapes Case Study Research

What Does Unit of Analysis Mean?

The unit of analysis identifies the entity or phenomenon to which the researcher's central analytical claims refer.

In case study research, the unit of analysis may be an individual, organization, program, event, institutional process, community, or another bounded system.

For example, a study investigating how a school implemented an adaptive learning program might define the school-level implementation process as its unit of analysis.

Teachers, students, administrators, and documents would contribute evidence about that process, but the principal findings would concern how implementation occurred within the school.

By contrast, a study examining individual teachers' professional decision-making might treat each teacher as a case, provided the design genuinely investigates each person's bounded experience or practice as a case.

The distinction depends on the analytical purpose rather than the surface characteristics of the dataset.

Is the Unit of Analysis Always the Same as the Case?

In many case-study designs, the primary unit of analysis corresponds closely to the case itself.

For example, if the study investigates one university as an organizational system, the university may be the case and primary unit of analysis.

However, case study methodology also permits embedded designs in which a larger case contains identifiable subunits that receive systematic analytical attention.

A university may be the overarching case, while its faculties or departments serve as embedded units.

Yin distinguishes holistic and embedded case-study designs partly through this analytical structure.

Researchers should therefore avoid assuming that every case study has only one analytical level. The primary case and its embedded units should be defined explicitly.

What Is the Difference Between the Unit of Analysis and the Unit of Observation?

The unit of observation is the entity or occurrence from which information is recorded. The unit of analysis is the entity or phenomenon about which the study seeks to make its principal analytical claims.

These may coincide, but they do not have to.

Suppose the researcher investigates the implementation of an AI assessment policy in one university.

The policy implementation process may be the unit of analysis. Individual faculty interviews, committee meetings, and documented policy revisions may constitute sources or units of observation.

However, the distinction requires care. An interview transcript is a data record, while the interviewee or interview event may be described as the observational unit depending on the methodological convention.

The researcher should specify what is being observed and what is being analyzed rather than treating these terms as interchangeable.

Can Participants Be Data Sources Without Being Cases?

Yes. This is one of the most important distinctions in case study methodology.

Suppose a researcher interviews thirty teachers about how a school introduced a new digital assessment policy.

If the research question concerns the school's policy implementation, the thirty teachers may be informants whose accounts help reconstruct the institutional process.

The study does not automatically contain thirty cases.

Alternatively, if the research investigates how each teacher individually negotiated professional identity and assessment practices, the teachers may constitute individual cases under an appropriately designed multiple-case approach.

Neither classification is determined by the number of interviews.

The researcher must identify what the analytical conclusions are intended to describe or explain.

How Does the Research Question Determine the Unit of Analysis?

A useful way to identify the unit of analysis is to ask what the research question is fundamentally about.

Consider the following questions:

These classifications are possibilities rather than universal prescriptions. A research question may be refined to focus on a different analytical level, which can change the unit of analysis.

For example, the university may serve as the broader setting while the policy-development process constitutes the actual case. Researchers should explain this distinction rather than automatically equating institution and case.

What Is a Holistic Case Study?

In Yin's typology, a holistic case-study design examines a case primarily as a single unit of analysis.

For example, a researcher might investigate the overall institutional response to the introduction of generative AI in assessment.

Faculty, administrators, students, and institutional documents may provide evidence, but the analysis focuses on the university's response as an integrated whole.

A holistic design can be appropriate when the research question concerns organizational processes or system-level explanations rather than systematic comparisons among identifiable subunits.

However, holistic analysis should not erase important internal differences. The researcher must still consider evidence that complicates the overall interpretation.

There is also a risk that an excessively broad case definition will produce superficial findings. The boundaries and analytical focus must remain clear.

What Is an Embedded Case Study?

An embedded case-study design includes identifiable subunits within a larger bounded case.

Suppose a university implements one AI assessment policy across four faculties. The researcher wants to understand both the university-wide implementation process and differences among faculties.

The university's policy implementation may constitute the overarching case, while the faculties serve as embedded analytical units.

The researcher may examine how each faculty interpreted the policy, allocated resources, and responded to academic integrity concerns.

The design permits analysis at more than one level, but it requires explicit connections between the embedded units and the overarching case.

A common danger is allowing the study to become a collection of unrelated faculty-level analyses without returning to the university-wide research question.

Embedded units should illuminate the case rather than unintentionally replace it.

How Is an Embedded Case Study Different From a Multiple-Case Study?

The distinction concerns the analytical relationship among the units being investigated.

The number of departments, participants, or sites alone does not determine which design is appropriate.

Researchers should examine whether the study requires one case with embedded units or several separate cases based on the research question and intended analytical contribution.

Can the Unit of Analysis Be a Process Rather Than a Person or Organization?

Yes. A case study may focus on a bounded process, provided the researcher defines its boundaries and explains why it constitutes a coherent object of inquiry.

Examples include the development of an institutional policy, the implementation of a curriculum reform, the resolution of a technology-related controversy, or the introduction of a new assessment system.

For example, the development of a university's AI governance policy between 2024 and 2026 could be defined as the case.

The university would provide the organizational context, while the policy-development process would be the primary unit of analysis.

This distinction may improve precision because the researcher is not claiming to investigate every aspect of the university.

The boundaries must still be specified. Researchers should clarify which decisions, actors, events, and periods belong to the process being investigated.

How Do Case Boundaries Relate to the Unit of Analysis?

The unit of analysis identifies what the study is about. Case boundaries specify the limits of that object of inquiry.

Suppose the unit of analysis is the implementation of a digital assessment policy.

The researcher might bound the case to one university, a defined implementation period, and the departments formally covered by the policy.

Decisions about earlier policy discussions, external regulators, alumni, and unrelated institutional initiatives would require explicit inclusion or exclusion criteria.

Without these boundaries, the case may expand indefinitely, making systematic data collection and interpretation difficult.

The researcher should therefore distinguish how case boundaries are established from the prior decision about what the unit of analysis actually is.

How Does the Unit of Analysis Affect Sampling?

Sampling in case study research can occur at several levels.

Researchers may select the case itself, choose embedded subunits, and recruit participants or identify documents that provide evidence about the case.

These decisions should not be collapsed into one sampling procedure.

For example, selecting a university because it offers a theoretically informative example of AI policy development is a case-selection decision.

Recruiting faculty members who participated in policy consultations is a participant-selection decision.

Selecting meeting minutes and policy drafts is an evidence-selection decision.

Each requires its own rationale. The researcher should explain how the selected sources contribute to understanding the defined unit of analysis.

Purposive selection may be appropriate, but convenience alone rarely provides a sufficient methodological justification for the case.

How Does the Unit of Analysis Affect Data Analysis?

The analytical strategy should preserve the relationship between the evidence and the unit of analysis.

Suppose the case is a university's policy-development process. Researchers may code interviews for themes such as faculty autonomy, academic integrity, and institutional accountability.

However, simply presenting themes across all interviews may not be sufficient to explain how the policy developed.

The analysis may also need to reconstruct decisions over time, compare participant accounts with documentary evidence, identify institutional turning points, and examine competing explanations.

If faculties are embedded units, the researcher may first examine each faculty's implementation experience and then analyze how those differences illuminate the university-wide case.

The purpose is to move from individual observations toward coherent claims about the defined case.

This does not require every case study to use the same coding or analytical procedure. The appropriate method depends on the research question and methodological tradition.

Can the Unit of Analysis Change During the Study?

Qualitative inquiry can involve refinement as researchers gain a better understanding of the phenomenon.

For example, an investigation initially framed around a university may reveal that the relevant case is a particular policy-development process rather than the institution as a whole.

Such refinement can be methodologically defensible when the researcher documents the reasoning and ensures that the revised questions, boundaries, sampling, and analysis remain coherent.

However, changing the unit of analysis after data collection may create gaps in the evidence or invalidate earlier sampling assumptions.

Researchers should distinguish legitimate conceptual refinement from changing the unit merely to accommodate available data or produce more attractive findings.

Major changes should be reported transparently, especially when the study was prospectively specified.

What Happens When Researchers Confuse Analytical Levels?

Confusing levels of analysis can produce conclusions that are not supported by the evidence.

For example, interviewing faculty members about their personal AI attitudes does not necessarily establish how the university as an organization makes policy decisions.

Individual perceptions may be relevant evidence, but organizational claims may require additional information about formal decisions, governance structures, and implementation processes.

Conversely, institutional policy documents do not automatically establish how every faculty member experiences or follows the policy.

Researchers should avoid moving between individual, departmental, and institutional claims without explaining how the evidence supports each level.

How Does the Unit of Analysis Affect Generalization?

The unit of analysis also determines the scope of the study's conclusions.

A case study of one university's AI policy implementation can develop an explanation of that institutional process. It cannot automatically estimate how frequently the same process occurs across all universities.

A multiple-case study may examine whether an explanation remains plausible across different institutions, but the cases are not necessarily a statistically representative sample.

Theoretical or analytical generalization may be possible when the study explains how findings relate to broader concepts or propositions.

Researchers should distinguish claims about the bounded case from claims about populations, other settings, or general causal mechanisms.

Understanding theoretical generalization in case study research helps prevent conclusions that exceed the analytical scope of the evidence.

Identifying the Unit of Analysis in a University AI Governance Study

Now suppose the research question changes to "How do thirty faculty members experience professional uncertainty when using generative AI?"

The original university-wide policy implementation process may no longer be the appropriate unit of analysis. A phenomenological or other qualitative design focused on individual experiences might better fit the revised question.

The example illustrates why the unit of analysis follows the analytical purpose, not the institution's name, the number of participants, or the researcher's initial methodological preference.

Common Mistakes When Defining the Unit of Analysis

How to Define the Unit of Analysis Before Collecting Data

Write your central research question and identify the entity or phenomenon about which the final conclusions will be made. Then determine whether that entity is a person, institution, program, event, or process.

Next, distinguish the case from the individuals and records that will provide evidence about it. If the design includes embedded units, specify their relationship to the overarching case.

A useful methodological statement should identify the primary case, explain its boundaries, specify any embedded units, and clarify how participants and documents contribute evidence.

This statement should remain consistent with the sampling procedures, analytical framework, and scope of the conclusions.

Before Finalizing the Unit of Analysis, Verify These Conditions

Frequently Asked Questions About the Unit of Analysis in Case Studies

The Unit of Analysis Determines What Your Case Study Can Legitimately Explain

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.