Aggregate A rate
78.0%
Simpson's Paradox Simulator
Edit subgroup counts and see how aggregate rates can reverse the within-group pattern.
At a glance
- Computes
- Subgroup rates and the pooled rate, showing when aggregation reverses the direction.
- You supply
- Successes and totals for two arms inside each of two subgroups.
- Use when
- You need to show how pooling unequal subgroups flips a comparison.
- Not for
- An association created by filtering the sample on a shared outcome. Berkson's Paradox Simulator
Aggregate B rate
82.6%
- Reversal status
- Reversal detected
- Group mix mechanism
- A has more hard cases
- Simulated reversal frequency
- Not run
| Group | A rate | B rate |
|---|
Value by category
Run the simulation to fill Monte Carlo metrics and convergence.
Reversal frequency against Trial
| Trial | A rate | B rate | Reversal frequency |
|---|
subgroup rate = successes / total; aggregate rate = summed successes / summed total
How?
How this is calculated
Simpson's paradox is flagged only when one arm has the higher rate in every subgroup and the other arm has the higher aggregate rate. Ties are not counted as a full reversal.
To model the same group structure as a graph, use the causal DAG builder and inspect which backdoor paths the group variable creates.
Formula: subgroup rate = successes / total; aggregate rate = summed successes / summed total
Sources
- Stanford Encyclopedia of Philosophy Simpson's Paradox. Stanford Encyclopedia of Philosophy. Retrieved .
Method last reviewed