Deviant Case Sampling — Selecting Extreme or Outlier Cases
Deviant Case Sampling · Also known as: extreme case sampling, outlier sampling, negative case sampling, deviant-case selection
Deviant case sampling is a purposive qualitative sampling strategy in which the researcher intentionally selects cases that are unusual, exceptional, or markedly different from the norm — outliers, extreme successes, or conspicuous failures. The goal is not statistical representation but deep learning from cases that illuminate the boundaries of a phenomenon, challenge prevailing assumptions, or reveal processes that typical cases obscure.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+1 more
When to use it
Use deviant case sampling when you want to understand the boundaries, failure modes, or exceptional successes of a phenomenon that typical cases cannot reveal — for example, an outlier program with dramatically better outcomes, or a process that catastrophically breaks down. It is appropriate when you have prior knowledge or baseline data sufficient to distinguish deviant from typical cases, and when the research question is exploratory or theory-refining rather than descriptive. Do not use it when the goal is to describe what is common or average in a population — random or stratified sampling is more appropriate for that; and do not use it as a default simply because a case is interesting or convenient.
Strengths & limitations
- Maximizes learning per case by concentrating on cases where outcomes are most pronounced and causes most visible.
- Efficiently challenges and refines existing theory by exposing boundary conditions and exceptions.
- Reveals mechanisms and processes that are obscured or confounded in typical, middle-range cases.
- Well-suited to applied research contexts where understanding failures or exceptional successes has immediate practical value.
- Complementary to other qualitative strategies — works well alongside maximum variation or typical case sampling in multi-strategy designs.
- Findings from deviant cases are not statistically generalizable; they offer theoretical, not population-level, insights.
- Identifying what is genuinely deviant requires reliable prior data or domain expertise — without this baseline, selection becomes arbitrary.
- Deviant cases may be deviant for idiosyncratic, context-specific reasons that do not transfer to broader theory.
- Small number of cases (often 2-6) limits the range of analytic comparison and may miss important variation within the deviant category.
Frequently asked
How many cases should I select?
There is no fixed rule. Deviant case sampling typically involves a small number of cases — often 2-6 — because each case is studied in depth. The goal is not numerical adequacy but conceptual richness. Select enough cases to allow meaningful comparison among deviant instances and to triangulate explanations, but keep the number manageable enough for thorough analysis.
What is the difference between deviant case sampling and maximum variation sampling?
Maximum variation sampling deliberately selects cases that span the full range of a dimension — from low to high — in order to document diversity and find common themes across diverse contexts. Deviant case sampling focuses specifically on the extreme ends of that range, or on cases that deviate from an established norm, in order to learn from the exception rather than to document variation. The intent and the analytic logic differ even though both may select some of the same cases.
Do I need quantitative data to identify deviant cases?
Not necessarily, though quantitative baseline data — survey results, administrative records, performance metrics — make identification more rigorous and defensible. In their absence, key-informant nominations, expert judgment, or systematic documentary evidence can be used to identify cases widely recognized as exceptional, provided the basis for that judgment is transparently documented.
Can deviant case sampling be combined with other sampling strategies?
Yes, and this is common in practice. A researcher might use maximum variation sampling to capture the broad landscape of a phenomenon, then add one or two deviant cases specifically to probe boundary conditions. The key is to document each sampling logic separately and to frame the analysis accordingly.
Is deviant case sampling only for qualitative research?
The term is predominantly used in qualitative and mixed-methods research, but the underlying logic — learning from outliers — applies in quantitative contexts too, for example in residual analysis or outlier investigation in regression. In quantitative work outliers are usually identified statistically; in qualitative work they are selected purposefully for in-depth study. The epistemological commitment to theory-refinement through exception is shared across both traditions.
Sources
- Patton, M. Q. (2002). Qualitative Research and Evaluation Methods (3rd ed.). Sage Publications. ISBN: 978-0761919711
- Flyvbjerg, B. (2006). Five misunderstandings about case-study research. Qualitative Inquiry, 12(2), 219-245. DOI: 10.1177/1077800405284363 ↗
How to cite this page
ScholarGate. (2026, June 3). Deviant Case Sampling. ScholarGate. https://scholargate.app/en/survey-methodology/deviant-case-sampling
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Case StudyQualitative↔ compare
- Maximum Variation SamplingSurvey Methodology↔ compare
- Purposive samplingSurvey Methodology↔ compare
- Snowball SamplingSurvey Methodology↔ compare
- Typical Case SamplingSurvey Methodology↔ compare