Confounding control · Analysis and reporting

Weight Concentration and Effective-Sample-Size Sensitivity

Four weights illustrate concentration and the effect of hypothetical caps on effective sample size.

Download SVG Open code and full worked example

When to use it

Use with overlap, balance and influence diagnostics; stability alone does not select a valid cap or causal target.

How to read it

Original weights 1,1,1,9 yield ESS 144/84 = 1.714. Capping changes the estimator and does not by itself define a new population.

Worked example

Four weights illustrate concentration and the effect of hypothetical caps on effective sample size.

Original weights 1,1,1,9 yield ESS 144/84 = 1.714. Capping changes the estimator and does not by itself define a new population.

Result: Use with overlap, balance and influence diagnostics; stability alone does not select a valid cap or causal target.

Produced by

Weight Truncation and Population Trimming Propensity Score Methods (PSM, IPTW)

Reference: Primary method sources are listed in the linked producing-method entries. The displayed data are original synthetic teaching examples, not published patient results.