Covariate Balancing
Updated 2026-08-10
INTRODUCTION
English translation pending.
CORE DEFINITION
A family of methods that adjusts for background variables affecting the outcome, so that treated and untreated groups have comparable distributions on those variables. Common techniques include propensity score matching, inverse probability weighting, and stratification. The key qualification is that balancing operates only on observed covariates: unmeasured confounders remain and can produce a spurious effect that no amount of balancing will remove. The quality of the result therefore depends as much on the plausibility of the unconfoundedness assumption as on the estimator.
SCAFFOLDING EFFECT
Reduce cognitive load
- Confounder listing: name the background variables that could affect both treatment and outcome. - Method choice: select matching, weighting, or stratification according to the data. - Balance check: verify standardized differences after adjustment rather than assuming it worked.
Anchor fast decisions
When treatment assignment depends on characteristics that also affect the outcome, the naive comparison mixes the effect of treatment with the effect of those characteristics. Balancing removes that mixture by making the groups comparable on the measured variables, so the remaining difference is more plausibly attributable to treatment. Because the procedure cannot see variables that were never measured, its validity rests on the assumption that the listed covariates capture the relevant differences.
MINIMUM ACTION
In progress 0/1Practice this model in one real situation:
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more
Source support: Explicit
- en.wikipedia.orghttps://en.wikipedia.org/wiki/Propensity_score_matchingverified
PRIVATE NOTES · Only visible to you
SAVED Q&A
ENTRY Q&A · Private saving available
Ask with a clear boundary
thinkingmodels answers from published entry context only.
Your question is sent to thinkingmodels. The answer uses public entry context only.
RELATED MODELS