Criminal justice statistics, like most statistics, rarely "speak for themselves," but rather require the user to supply a proper context to understand and interpret the numbers. For these tables, here are some of the relevant issues.

Where do the data come from? The population figures are for 1999, and were estimates from the U. S. Census Bureau, while the arrest statistics are from the Justice Statistics Center at Minnesota Planning. Both breakdown the numbers by racial group. Do the two sources use the same methodology to determine who, for example, is an American Indian? It is possible that the Census method, probably based on self-report of individuals, results in more people assigned to this group than the state's method, which may be based on the judgment of the arresting officer.

The arrest statistics are number of arrests, not number of people arrested. One person arrested several times would be counted several times.

This is not experimental data, but rather an observational study. Any observed differences in rates could be due to other, uncontrolled, factors. For example, if the age or income distributions of racial groups are different, then differences in rates may be due to age or income differences, not racial differences.

Is the rate arrests per 1000 population in a group a relevant number? Would not a more relevant number be the number of arrests per 1000 offences in the group be better? Of course, this latter is not computable because we don't know how many offenses were committed.

An interesting article that addresses some of the issues in understanding these numbers, but in regard to arrest statistics for NFL football players, is "Criminal violence of NFL players," which appeared in Chance Magazine in Summer, 1999. You can download a copy from http://www.public.iastate.edu/~chance99 .