Evidence-appraisal glossary
Positivity
A condition for valid causal comparison requiring that every type of person in the study has a nonzero chance of receiving each treatment being compared. If some group could only ever receive one treatment, its effect there cannot be estimated from the data.
Also called: experimental treatment assignment.
Positivity, also called the experimental treatment assignment assumption, means there are both treated and untreated people at every combination of the confounders you adjust for. When it fails, methods such as weighting must extrapolate into regions where no real data exist, producing unstable or misleading estimates. Near-violations show up as extreme propensity scores or enormous weights, which is why analysts check the overlap between groups before trusting an adjusted result.
This is a plain-language methodology definition for reading research. It is general education, not medical advice.