Extreme and Deviant Case Sampling

Description

Deliberately recruiting outliers, the most frustrated, most expert or unexpectedly successful users, to expose mechanisms and boundary conditions typical cases hide.

Typical cases show average experience; extreme cases show where systems break hardest or work unexpectedly well. Extreme case sampling selects participants at the tails of a relevant distribution, often identified from analytics, support data or survey scores. Deviant case sampling targets those who do not fit the emerging explanation, which makes them the strongest test of a theory. Positive deviance, from public health, studies people who succeed despite the same constraints as others, revealing workable strategies. Pairing extremes with typical cases sharpens comparison. Rigour lies in defining 'extreme' from the research question, using data to locate cases, and treating outliers as scope conditions rather than noise.

Attributes

Loss type
Latent
Stage
Discovery
Effort
Days

Steps

  1. Identify what counts as extreme for your research question: most experienced, most frustrated, fastest adopter, most resistant.
  2. Analytic pair sampling: Pair extreme case sampling with typical case sampling. The comparison generates the most insight.
  3. Use quantitative data (usage analytics, NPS scores) to identify participants at the extremes of distributions. Recruit from those extremes purposively.
  4. Do not treat extreme cases as errors. Treat them as boundary conditions revealing scope conditions of your emerging theory.

Sources

  • Sampling and Choosing Cases in Qualitative Research
  • Sage Handbook of Qualitative Data Analysis (Flick)
  • The Power of Positive Deviance (Pascale, Sternin & Sternin)
  • Qualitative Research and Evaluation Methods (Patton)
  • Miles, M. & Huberman, A. M., Qualitative Data Analysis: An Expanded Sourcebook, 2nd ed. (1994)

Also known as

  • Outlier sampling
  • Deviant case sampling
  • Positive deviance sampling
  • Analytic pair sampling
  • Bright-spot analysis

Categories

  • Research Operations
  • Research Design & Rigor