25% variation in the measurement of murders

A great example of the importance of definitions, the importance of apples-to-apples comparisons and the importance of measurement margin awareness.

From How Many Homicides were there in 2010? by Alex Tabarrok.

How many homicides were there in 2010 in the United States? Well, that’s easy. Let’s just do some Googling:

1.12,966, FBI, Crime in the United States 2010.
2.13,164, FBI, Crime in the United States 2011 (2010 figure).
3.14,720, Bureau of Justice Statistics (Table 1, based on FBI, Supplementary Homicide Statistics).
4.16,259, CDC (based on death certificates in the National Vital Statistics System).

Between the smallest and largest figures there is a difference of 3,292 deaths or 25%!

What explains the differences? The explanations are incomplete, but according to Tabarrok the differences are broadly due to.

1. The original measure
2. The updated measure with late calendar returns
3. Includes reports of murder rather than just victim count.
4. Includes justifiable homicides (death inflicted in self-defense.)

When discussing crime, murder statistics are often resorted to as a bedrock of crime indication. Unlike fraud, burglary, robberies, etc. most murders are known about relatively quickly. It is hard for someone to disappear without being noticed. Murders are given a high priority in terms of investigation and so there is a higher clearance rate than most other crimes.

All that being true, the above numbers demonstrate that it is still a relatively murky business and care should be exercised. 25% is a really large measurement margin.

Based on these numbers, between 4-5% of deaths are "justified" in the sense of being some form of self-defense from a legal perspective.