Causality & Research.

Balance the groups and separate a treatment effect from a background advantage.

Use encouragement to identify an effect, then deliberately break the instrument.

Inspect a policy cutoff and discover why a nearby second policy matters.

Compare changes, then see what happens when the comparison trend is wrong.

Search thirty hypotheses without mistaking every small p-value for a discovery.

Add noise to the predictor and investigate why a fitted slope can look too weak.

Build repeated confidence intervals and reveal which contain the generating mean.

Change sample size and effect size, then compare expected power with a hundred simulated tests.

Observe a coin stream and stop only when a predeclared likelihood-ratio boundary is crossed.

Inspect resampled means and see what happens when the largest observed value is omitted.

Plot four datasets and omit individual observations to inspect the structure behind a regression.

Compare an online-only survey frame with a complete fictional population.

Open a file drawer of studies hidden by a significance-only publication rule.

Inspect individual cells before translating a district-level association into a personal claim.

Narrow the selected range of one variable without changing its generating relationship.

Move one observation and compare mean, median and trimmed mean.

Compare rival predicates that fit every past observation but project different future colours.