03 · What You Need to Know
How to Use Extreme and Unusual Cases Without Overstating Their Evidence
What Is an Extreme or Unusual Case?
An extreme case is one that lies toward an exceptional end of a distribution or differs markedly from other cases on a characteristic relevant to the research question.
For example, a university might have an exceptionally high level of faculty adoption of AI-assisted assessment relative to comparable universities.
An unusual case may instead be distinctive because of its institutional configuration, history, governance arrangements, or combination of circumstances.
These concepts overlap, but they are not always identical. A case can be statistically extreme on a measured variable without being unusual in every other respect. Conversely, a case can exhibit a rare combination of characteristics without having the highest or lowest value on any single indicator.
Researchers should therefore explain the particular dimension on which the case is considered exceptional.
How Is an Extreme Case Different From a Deviant Case?
The terms extreme, unusual, and deviant are sometimes used loosely, but they can describe different selection logics.
An extreme case has an unusually high, low, or otherwise exceptional value on a relevant characteristic.
A deviant case, in an important methodological usage, is one whose observed outcome differs substantially from what a theoretical model or established empirical pattern would predict.
For example, a university with exceptionally high AI adoption may be an extreme case. A university with limited infrastructure that nevertheless achieves much higher adoption than a well-supported theoretical model predicts may be a deviant case.
The distinction matters because deviant-case investigation often seeks to understand why an existing explanation does not adequately account for the observed result.
Seawright and Gerring's work on case-selection techniques provides a useful basis for distinguishing these selection strategies rather than treating all atypical cases as methodologically equivalent.
When Can an Extreme Case Reveal Otherwise Hidden Processes?
Extreme cases may make certain processes particularly visible because the phenomenon is unusually intense, prolonged, or consequential.
For example, a school experiencing severe and persistent connectivity disruptions may reveal how teachers adapt instructional practices under technological constraints.
Such adaptations might be less apparent in schools where connectivity problems are occasional or easily resolved.
The case may therefore provide rich evidence about institutional responses, informal practices, and decision-making under pressure.
However, greater visibility does not guarantee that the same processes operate identically under ordinary conditions.
The researcher should distinguish what the case reveals directly from broader propositions that require further examination.
When Is an Exceptionally Successful Case Worth Studying?
An exceptionally successful case may be informative when the researcher wants to understand how a particular achievement developed, what conditions supported it, or whether existing explanations adequately account for it.
Suppose a university achieves unusually sustained adoption of an AI-assisted tutoring system while comparable institutions struggle.
A case study could investigate leadership decisions, faculty involvement, technical support, student participation, and implementation history.
The research might identify plausible mechanisms that deserve examination elsewhere.
However, studying success alone does not establish that the observed practices caused success. The same practices may be present in unsuccessful institutions, while unobserved conditions may have contributed to the exceptional result.
Researchers should therefore avoid presenting a successful case as a universal model of best practice without comparative or other supporting evidence.
When Is an Exceptionally Unsuccessful Case Useful?
Unsuccessful cases can reveal implementation difficulties, institutional constraints, or unintended consequences that successful cases may conceal.
For example, a university may discontinue a costly learning analytics platform after repeated concerns about data quality, faculty workload, and student privacy.
Investigating this case could clarify how technical limitations interacted with institutional governance and professional practice.
Such findings may be valuable for understanding potential failure mechanisms and informing future implementation decisions.
However, the researcher should avoid assuming that every observed difficulty caused the program's failure.
Chronological evidence, competing explanations, and attention to institutional context are needed to develop a credible account.
How Do You Establish That a Case Is Genuinely Extreme?
Researchers should identify the reference population and the characteristic used to establish unusualness.
For example, describing a university as having exceptionally high AI adoption requires a defensible basis for comparison.
Possible evidence includes administrative records, comparable institutional surveys, implementation reports, or other systematically collected information.
The researcher should examine whether the measures are defined consistently across cases and whether the reference population is appropriate.
A university may have the highest adoption rate among five conveniently observed institutions without being exceptional across the wider sector.
When reliable comparative data are unavailable, researchers should qualify the claim and explain the specific evidence supporting the case's distinctiveness.
Does an Extreme Case Need to Be a Statistical Outlier?
No. Extreme-case selection may be informed by quantitative distributions, but it does not universally require a formal statistical outlier test.
For example, a researcher might select a school because it experienced an unusually prolonged disruption to digital learning, even when no comprehensive dataset of disruption durations exists.
The unusualness could be established through documented institutional history and comparison with known practices.
Nevertheless, a purely impressionistic claim that a case is exceptional may be difficult to defend.
The researcher should provide a transparent explanation of the selection criteria and the evidence supporting them.
Why Can Selection on the Outcome Be Problematic?
Selection on the outcome occurs when cases are chosen because they exhibit a particular result, such as exceptionally high achievement or unusually poor implementation.
This can be useful for understanding how an observed result developed, but it creates limitations for causal inference.
Suppose researchers select only universities where AI tutoring adoption was highly successful and identify extensive faculty training as a common feature.
Without examining universities where similar training occurred but adoption remained weak, the researchers cannot establish that training explains the difference in outcomes.
The problem is especially consequential when the study attempts to identify conditions that distinguish success from failure.
Outcome-selected cases can contribute to causal-process investigation, but conclusions must reflect the limited variation and the possibility of competing explanations.
Can Regression to the Mean Affect Extreme-Case Research?
Yes, particularly when cases are selected because of unusually high or low measurements.
Regression to the mean describes the tendency for subsequent measurements to be less extreme when initial observations contain random variation.
For example, a school selected because of an exceptionally low examination score may show improvement the following year even without a meaningful intervention.
If researchers interpret the improvement as evidence of successful institutional reform, they may overstate the reform's contribution.
Historical outcome records, repeated measurements, and appropriate comparisons can help assess whether the initial extreme value reflected a persistent condition or a temporary fluctuation.
However, these measures do not automatically establish a causal effect.
How Does Extreme-Case Selection Differ From Typical-Case Selection?
Extreme-case selection emphasizes exceptional characteristics or outcomes, while typical-case selection focuses on cases that reflect relatively ordinary conditions within a defined reference population.
| Dimension |
Extreme or unusual case |
Typical case |
| Selection rationale |
Exceptional characteristics, outcomes, or circumstances |
Characteristics considered relatively ordinary |
| Potential contribution |
Investigating unusual mechanisms, anomalies, or intensified processes |
Understanding processes under more common conditions |
| Principal interpretive concern |
Overgeneralization and outcome-based selection |
Unsupported assumptions of representativeness |
| Typical research interest |
Why did this exceptional situation develop? |
How does this phenomenon operate under ordinary conditions? |
Neither approach is inherently stronger. Researchers should choose according to the question and the analytical opportunity provided by the case.
The distinction is examined further in selecting a typical case for case study research.
Can an Extreme Case Be a Critical Case?
Potentially, but extreme and critical describe different selection rationales.
An extreme case is selected because of its exceptional characteristics. A critical case is selected because its conditions make it particularly informative for evaluating a theoretical proposition.
For example, a university with exceptionally high technological resources might be a critical case for examining a proposition that inadequate infrastructure is the principal barrier to AI adoption.
If implementation still fails under highly favorable infrastructure conditions, the finding may challenge a simple version of that explanation.
However, the critical-case argument must be established through theory and context. Being extreme does not automatically make a case critical.
How Can an Extreme Case Contribute to Theory?
Extreme cases may expose limitations in existing explanations, reveal previously overlooked processes, or suggest conditions under which theoretical propositions require modification.
For example, an institution that achieves substantial AI adoption despite limited infrastructure may challenge an explanation that treats infrastructure as the dominant determinant of adoption.
Closer investigation might reveal compensating organizational practices, external partnerships, or alternative forms of access.
These observations could motivate theoretical refinement, but their broader applicability would require further investigation.
The contribution is better understood as analytical or theoretical rather than automatically statistical.
Researchers should explain how case evidence relates to broader theoretical propositions without assuming that the unusual case represents an entire population.
Should You Study One Extreme Case or Several?
A single extreme case may be appropriate when the research seeks intensive understanding of a distinctive process and substantial evidence is available.
Several extreme cases may be useful when researchers want to compare different forms of exceptional performance, failure, or institutional response.
For example, comparing unusually successful and unusually unsuccessful AI implementations may help examine contrasting conditions.
However, the comparison requires explicit selection logic and attention to other differences between cases.
Adding cases does not automatically eliminate selection bias or establish causal relationships.
The decision should follow the principles of choosing between single-case and multiple-case study designs.
What Evidence Is Particularly Important in an Extreme Case?
Extreme cases often attract compelling narratives, which makes systematic examination of contradictory evidence especially important.
Researchers should reconstruct relevant events, document the basis for the case's unusualness, and examine alternative explanations.
Interviews may reveal how participants interpret the exceptional outcome, while institutional records and observations may help establish what actually occurred.
Evidence that contradicts the preferred explanation should be considered rather than dismissed as an exception.
Researchers should also distinguish unusual conditions that preceded the outcome from changes that emerged as consequences of it.
These practices help prevent an attractive case narrative from being mistaken for a demonstrated causal explanation.
When Is an Extreme Case the Wrong Choice?
An extreme case may be unsuitable when the research primarily seeks to understand ordinary implementation conditions or estimate how frequently a phenomenon occurs.
For example, studying a university with exceptionally high AI adoption may provide limited direct information about the implementation challenges faced by institutions with average resources and ordinary governance arrangements.
Extreme-case selection may also be weak when the exceptional status is based on unreliable measurements or when the researcher cannot access evidence necessary to investigate the phenomenon.
If the objective is to estimate population patterns, a suitably designed survey or quantitative study may be more appropriate.
Researchers should first establish a defensible case-selection rationale before choosing a case because its outcome is striking.
