Defunding Police — An exploratory data analysis

There is a growing pressure on police departments and politicians in the United States to do something about the increased rate of police brutality. Discussions to address police brutality include ideas from reforming law enforcement tactics to defunding or complete abolishment of the police department.

The purpose of this analysis to understand if defunding police reduces or increases crime rates?

I analyzed crime and expense data from Uniform Crime Reporting Statistics (UCR) and State & Local finance data from the Urban Institute for the period 2000 to 2014. I studied the impact of state & local expenses on correction, police, education, health, hospital, housing, public welfare, and unemployment on the crime rate.

A quick review of the data shows no correlation between police expenses and crime rates. The most significant correlation is a negative correlation between public welfare expanses and crime rates.

Fig 1 — Correlation

The average crime rate in the United States is 3671 crimes per 100000, and the average police & correctional expense rate is $57,727 per 100000. Figure 2 below shows the data broken down by the various regions and figure 3 by states.

Fig 2 — Crime vs. Expense Rates by Region

Fig 3— Crime & Expense Rate by State

The standard linear regression model indicated that an increase in police expenses resulted in an increase in crime rates, provided all other predictor variables remain constant.

Fig 4 — Coefficients to scale for Full Model

I then performed cross validation, with n-folds = 100, with the full model, stepwise model, ridge model, and lasso model. All the models indicated that an increase in police expenses resulted in an increase in crime rates, provided all other predictor variables remain constant.

Fig 5 — Coefficients of predictor variables.

Based on the preliminary analysis of the data, I identified that an increased expense for education, public welfare and unemployment would reduce the crime rates.