By now everyone probably has heard of the 13/50 meme or any of its variations and it's usually acompanied with something along the lines of
>13 percent of the population, yet 50% of the murders, they are overrepresented by almost 4 times!

The thing is that just dividing the incidences by the population size does not seem to be a particularly good metric.
For example 50% of the population committing 100% of the murder would be an even bigger discrepancy, but would only be registered as two times the representation by this metric.
Is there any better way to measure overrepresentation?