Skip to content
Omnikalc

Simpson's paradox calculator: check whether pooling the data reverses the result

Enter successes and cases for two options in two subgroups to compare the subgroup rates with the overall rates and see whether the trend reverses.

Updated on
Your inputs

Value out of range

Value out of range

Value out of range

Value out of range

Value out of range

Value out of range

Value out of range

Value out of range

Results

Simpson's paradox present (1 = yes, 0 = no)

1

Paradox: one option wins in both subgroups but loses overall
Success rate of A in subgroup 1
93.1 %
Success rate of B in subgroup 1
86.7 %
Success rate of A in subgroup 2
73.0 %
Success rate of B in subgroup 2
68.8 %
Overall success rate of A
78.0 %
Overall success rate of B
82.6 %

Averages can lie when groups are mixed. The Simpson’s paradox calculator takes the successes and cases of two options in two subgroups and gives the success rate of each option in each subgroup, the overall rate and a flag that tells you whether the comparison reverses once the data is pooled.

How the calculation works

  • Enter the successes and the number of cases for two options, A and B, in each of two subgroups.
  • The calculator finds the success rate of each option within each subgroup, and then the overall rate after the subgroups are pooled.
  • Simpson’s paradox appears when one option has the higher rate in both subgroups but the lower rate overall, or the reverse.
  • The reversal happens because the subgroups are different in size and in difficulty, so pooling weights the options differently.

Example

In a classic study of kidney stone treatments, treatment A succeeded in 81 of 87 small stones and 192 of 263 large ones, while treatment B succeeded in 234 of 270 small stones and 55 of 80 large ones. A wins in both groups (93.1 % against 86.7 %, and 73.0 % against 68.8 %), yet overall A reaches 78.0 % and B reaches 82.6 %.

Why it matters

The paradox shows up in medicine, admissions, sports statistics and business reports, and it is a standard lesson in statistics courses. Checking the numbers both ways is a good habit before drawing a conclusion.

Good to know

The calculator only compares two options in two subgroups and cannot tell you which view is causally right. That needs knowledge of how the data was produced. A reversal also depends on exact counts, so small changes in the data can remove it.

The Bayes theorem calculator shows another way that intuition about rates can fail, and the weighted average calculator explains how pooling weights groups by size.

Frequently asked questions

What is Simpson's paradox?

It is a situation in which a trend that appears in each of several groups reverses or disappears when the groups are combined. It is named after Edward Simpson, who described it in 1951, although earlier statisticians had noticed it.

Why does it happen?

It happens when the groups differ in size and the option is not spread evenly across them. In the kidney stone example, treatment A was used mostly on the harder large stones, so its overall rate was pulled down even though it did better in each type.

Which result should I trust, the subgroup or the overall rate?

It depends on the cause. If the subgroup is a confounding factor that affects both the choice and the outcome, such as the severity of the case, the subgroup comparison is usually the fairer one. If the subgroup is a result of the treatment itself, the overall rate may be the better guide.

How can I avoid being misled?

Look for hidden factors that could split the data, compare like with like, and be careful with summary percentages when group sizes differ a lot. Randomized assignment prevents most of the problem.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.