Description
Compares users who experienced a specific harm with matched users who did not, measuring prior exposure to design features, efficient for rare or severe outcomes.
A case-control study starts from outcomes: cases who experienced the outcome are compared with controls who did not, and prior exposures are measured. The treatise proposes UX case-control studies comparing users who suffered a specific harm (financial loss from a dark pattern, task failure from an accessibility barrier, medical error from a clinical interface) with matched users who did not, measuring prior exposure to design features. They are particularly valuable for rare or severe outcomes (clinical UX errors, fraud-related design harms, automotive interface accidents) where prospective cohorts would need impractically large samples. The odds ratio estimates association; rigour depends on control selection from the same source population, matching and avoiding recall bias by using logs.
Attributes
- Discipline
- Epidemiology & Population Health
- Loss type
- Unmigrated
- Stage
- Post-launch
- Data
- Quantitative
- Effort
- Weeks
Steps
- Define cases precisely and identify them (incident reports, complaints, logs).
- Select controls from the same source population, matched on key factors (for example time, product version).
- Measure prior exposure to candidate design features, preferably from logs to avoid recall bias.
- Compute odds ratios with conditional logistic regression for matched designs.
- Adjust for confounders and test dose-response.
- Interpret with Bradford Hill criteria and confirm with experiments where possible.
Sources
- Lost Methods in UX — A Complete Treatise (Gopinath)
- Schlesselman JJ (1982) Case-Control Studies: Design, Conduct, Analysis
- Doll R & Hill AB (1950) Smoking and carcinoma of the lung. BMJ
- Rothman, Greenland & Lash, Modern Epidemiology (3rd ed., 2008)
- Doll & Hill, Smoking and carcinoma of the lung (1950), BMJ
- Schlesselman, Case-Control Studies (1982)
Also known as
- Case-control design
- Retrospective exposure comparison
- Case-control study
- Retrospective case-control
- Outcome-based sampling
- Case-Control Study Design
Categories
- Epidemiology & Population Health