03 · What You Need to Know
How to Build a Defensible Case-Selection Strategy
Why Does Case Selection Matter?
Case selection shapes what the researcher can observe, which explanations can be examined, and what conclusions the evidence may support.
A case with extensive records may allow detailed reconstruction of institutional processes. A case with an unusual outcome may reveal mechanisms that are difficult to observe elsewhere. A typical case may help researchers understand how a phenomenon operates under relatively ordinary conditions.
However, each selection creates different inferential opportunities and limitations.
For example, selecting a university because it successfully implemented an AI policy may help researchers investigate how success unfolded. It does not automatically establish that the identified practices caused success or that unsuccessful institutions lacked those practices.
Researchers should therefore treat case selection as part of the research design rather than merely a logistical step preceding data collection.
What Is the Difference Between Convenience and Purposive Case Selection?
Convenience selection prioritizes accessibility, proximity, existing relationships, or ease of recruitment.
Purposive selection identifies cases because their characteristics are relevant to the research question and analytical purpose.
For example, selecting a university because the researcher works there is primarily a convenience rationale.
Selecting the same university because it has a documented, unusually transparent AI governance process that allows examination of competing policy decisions may provide a purposive rationale.
The two are not necessarily incompatible. A theoretically informative case may also be accessible.
However, the researcher should explain the case's analytical value independently of its convenience and consider whether the researcher's relationship to the setting creates additional ethical or interpretive concerns.
What Is an Information-Rich Case?
An information-rich case provides substantial opportunities to investigate the phenomenon of interest.
Patton's purposive sampling framework emphasizes selecting cases that can yield meaningful information relevant to the research purpose.
For example, a university that preserved policy drafts, consultation records, implementation reports, and stakeholder communications may offer unusually detailed evidence about AI policy development.
However, information richness is not equivalent to simply having many documents or participants.
The available evidence must be relevant, credible, and sufficiently varied to support the intended analysis.
A case may be well documented but analytically unsuitable if its circumstances do not address the research question.
When Should You Select a Typical Case?
A typical case is selected because it reflects characteristics considered relatively ordinary or common within a defined reference population.
For example, a researcher might investigate AI policy implementation in a medium-sized university with governance arrangements, resources, and academic programs similar to those of many institutions in a specified sector.
The purpose may be to understand how a process unfolds under comparatively ordinary conditions rather than in an exceptional setting.
However, typicality must be established relative to a clearly defined population and relevant dimensions.
A university may be typical in enrollment size but unusual in technology infrastructure or governance.
Researchers should therefore explain what makes a selected case typical rather than treating typical as a synonym for familiar or average.
When Should You Select an Extreme or Unusual Case?
An extreme or unusual case is selected because it differs substantially from ordinary cases on a characteristic relevant to the research question.
For example, a university may have achieved exceptionally rapid adoption of AI-assisted assessment despite limited technical infrastructure.
Investigating such a case could reveal institutional practices, adaptations, or contextual conditions that are less visible in more ordinary settings.
However, selecting a case because of an extreme outcome creates interpretive risks.
The researcher may overemphasize explanations consistent with the unusual result or overlook the role of chance, historical circumstances, and unobserved factors.
Researchers should distinguish the analytical value of an extreme case from unsupported claims that its distinctive characteristics caused the observed outcome.
What Is a Critical Case?
A critical case is selected because its characteristics provide a particularly informative opportunity to examine a proposition or theoretical expectation.
For example, researchers may propose that faculty consultation is necessary for successful AI assessment policy implementation.
A university with unusually extensive consultation but persistent implementation resistance could provide a challenging case for that proposition.
However, describing a case as critical requires more than asserting that it is important.
The researcher must explain the theoretical logic that makes the case especially informative and specify what different findings would imply for the proposition.
Case selection should be guided by that logic before the researcher interprets the results.
What Is a Revelatory Case?
A revelatory case provides access to a phenomenon that has previously been difficult for researchers to observe.
For example, a university may permit researchers to examine confidentially managed policy-development processes, subject to appropriate permissions and safeguards.
Such access could make it possible to investigate how institutional decisions actually unfolded rather than relying exclusively on public policy documents.
The case's value lies in the opportunity to observe a previously inaccessible process.
However, privileged access should not be confused with unrestricted ethical permission. Researchers must protect confidential information, manage conflicts of interest, and comply with applicable institutional requirements.
The revelatory rationale should explain what new analytical access the case provides.
What Is a Longitudinal Case?
A longitudinal case is investigated across meaningful periods to understand development, continuity, or change.
For example, a researcher may examine how a university's AI governance policy evolved over four academic years.
The case may provide opportunities to investigate changing stakeholder relationships, implementation problems, policy revisions, and institutional learning.
Longitudinal selection is especially appropriate when temporal development is central to the research question.
However, the researcher must define the relevant periods and consider whether historical records are sufficiently comparable.
A long observation period does not automatically provide a stronger explanation if important events and evidence are missing.
How Do You Select Cases for Theoretical Variation?
Cases may be selected because they differ on characteristics expected to influence the phenomenon.
For example, researchers investigating AI governance might select universities with centralized, decentralized, and collaborative policy-development arrangements.
The variation provides opportunities to examine whether governance conditions are associated with different implementation processes.
However, the selected dimensions should be theoretically justified rather than chosen simply because they are easy to observe.
Researchers should also consider whether other characteristics differ systematically across cases, potentially complicating interpretation.
Comparative case selection can strengthen analytical inquiry, but it does not automatically establish controlled causal comparison.
How Does Case Selection Differ Between Single-Case and Multiple-Case Studies?
In a single-case study, the researcher must justify why the selected case is sufficiently informative to warrant intensive investigation.
In a multiple-case study, the researcher must justify both the selection of individual cases and the logic connecting them.
Yin's replication logic, for example, distinguishes cases expected to produce similar findings from those expected to produce theoretically meaningful contrasts.
Other comparative traditions may emphasize contextual variation, interpretive comparison, or theoretically relevant configurations.
The decision between single-case and multiple-case study designs should therefore follow the analytical purpose rather than a belief that more cases necessarily produce greater rigor.
Should You Select Cases Based on the Outcome?
Outcome-based selection can be useful for some research purposes, but it creates important inferential limitations.
For example, selecting a university because its AI policy implementation was unusually successful may be appropriate for investigating how that success was experienced and organized.
However, examining only successful institutions cannot establish that a particular governance practice causes success.
The practice may also occur in unsuccessful institutions, or the observed success may depend on other conditions.
Selection on the dependent variable is particularly problematic when the research intends to compare causal explanations but includes little or no variation in the outcome.
Researchers should distinguish case selection for descriptive or process-oriented purposes from selection intended to support broader causal claims.
Can Researchers Select Cases Using Preliminary Data?
Yes. Preliminary data can help identify cases that meet theoretically relevant criteria.
For example, an initial survey of universities may reveal differences in AI governance structures, implementation maturity, or faculty participation.
The researcher could then purposively select cases representing meaningful configurations.
However, the selection criteria should be documented, and researchers should distinguish exploratory selection from confirmatory evaluation.
When cases are chosen after examining outcomes, the implications for causal interpretation and selective reporting should be acknowledged.
Preliminary evidence should improve the transparency of selection rather than provide an opportunity to choose only cases that support a preferred explanation.
How Do You Construct a Case-Selection Matrix?
A case-selection matrix can make the rationale more explicit by comparing candidate cases on characteristics relevant to the research question.
| Selection criterion |
What to examine |
Why it matters |
| Relevance |
Whether the case exhibits the phenomenon of interest |
Ensures the case can answer the research question |
| Theoretical significance |
Whether the case can inform a proposition or broader issue |
Supports analytical contribution |
| Contextual characteristics |
Institutional size, governance, resources, or history |
Clarifies typicality, variation, or distinctiveness |
| Evidence availability |
Access to relevant documents, participants, and observations |
Supports adequate investigation |
| Feasibility |
Permissions, time, resources, and ethical constraints |
Determines whether the study can be completed responsibly |
| Inferential limitations |
Outcome selection, unusual conditions, or restricted access |
Clarifies what conclusions may not be justified |
The matrix should guide judgment rather than produce a mechanical score that automatically determines the best case.
Criteria should be weighted conceptually according to the research purpose, and researchers should explain consequential trade-offs.
What If the Most Informative Case Is Not Accessible?
Case selection must balance analytical value with feasibility.
A theoretically compelling case may be unsuitable if researchers cannot access the records or participants necessary to investigate it adequately.
However, selecting the most accessible alternative should not be treated as methodologically equivalent without further justification.
Researchers may need to refine the research question, select a different theoretically informative case, or acknowledge that access constraints limit the intended contribution.
Where access is uneven, the methods section should explain how the limitations affect evidence adequacy and interpretation.
How Can Researcher Familiarity Affect Case Selection?
Researchers sometimes study institutions where they work or have established professional relationships.
Such familiarity can improve access and contextual understanding, but it may also influence participant recruitment, interpretation, and the willingness of participants to disclose sensitive information.
For example, faculty members may respond differently when interviewed by a colleague involved in institutional AI governance.
Researchers should consider positionality, conflicts of interest, confidentiality, and power relationships.
Reflexive documentation and transparent reporting can help readers understand how researcher involvement may have shaped the investigation.
Familiarity is not inherently disqualifying, but it should not substitute for a case-selection rationale.
How Should Case Selection Be Reported?
A methods section should explain the selection pool, relevant eligibility criteria, the rationale for choosing the case, and the relationship between selection and the research question.
Researchers should identify whether the case was selected as typical, extreme, critical, revelatory, longitudinal, or otherwise theoretically informative.
They should also describe relevant access constraints and acknowledge how the selection affects the interpretation of findings.
For multiple-case studies, the rationale should explain why the selected cases are examined together and how their similarities or contrasts contribute to the inquiry.
The strongest justification is specific enough that readers can understand why this case was selected instead of plausible alternatives.
