In an attempt to test my stress model, I studied Colorado crime rates. I immediately noticed that violent crime rates dipped below the minimum limit predicted by the model, appearing to disprove my most basic assumption: that people in the outer edges of a population (and presumably under the most stress) were responsible for violent crime.
The close parallel between murder rates and average stress suggested that violent crime might behave the same way. I revamped the model, this time based on the premise that violent crime is proportional to average stress. The proportionality factor became immediately obvious: Normalized stress (average stress as a fraction of the maximum possible average stress) is the per capita violent crime rate multiplied by the number of years in an average lifetime. While my basic conclusions are left unchanged, the new model allows any violent crime rate between zero and 1,299 (per 100 thousand people per year).
Of practical concern is the huge question of what real-world variables correspond to “adjusted power” that can be measured and manipulated in a quantifiable way.