Death penalty example of Simpson's paradox.
Data from Radelet (1981)




Defendants         Victims          Death Penalty   Percentage yes
Race               Race
                                    YES      NO 
__________________________________________________________________

White              White            19       132     12.6
                   Black             0         9      0.0



Black              White            11        52     17.5 
                   Black            6         97      5.8 

__________________________________________________________________

Note.

1. If victim is white, blacks get death penalty more often
   (17.5 vs 12.6%)
 
2. If victim is black, blacks get death penalty more often
   (5.8 vs 0.0%)

3. Ignoring victims race, 19/160 white defendents got the
   death penalty, whereas 17/166 black defendants got the
   death penalty. Whit's got the death penalty more frequently 
   than blacks.

4. Conclusion (among others) partial association is not the same
as marginal association.

5. Practical conclusion, when someone presents you with results always
ask if there were other important variables that were ommitted 
from the analysis. Their inclusion could change the conclusions.