Chapter 4 of 10All chapters
Chapter 4 of 10
Missing denominators
Counts against rates.
Per what
Comparing raw counts across groups of different size is meaningless. Population, exposure or opportunity has to be in the denominator.
- More accidents in a large city is not a higher accident rate.
- Percentage changes need the base: a 100 percent rise from two cases is four.
Sample size
Survey percentages without the sample size hide how much noise is present, and the subgroups that make the best headlines are usually the smallest.