Aug. 11, 2001 (©2001)
County-by-county concealed carry license prevalence, calculated from census bureau population data and concealed carry license data available from the Florida Attorney General's (AG's) web site, was regressed against county-by-county 1997 "violent" (against person) and "nonviolent" (property) crime rates from the same site and against county-by-county data from the Census Bureau on '95 median household income, '97 minority proportion, '97 population density (logarithmic, to base 10), and '95 population portion "in poverty."
Firearm concealed carry license suspension rates for the same period as the license issuance rates, calculated from data also from the AG's web site, were regressed against the same factors and also against the '97 crime rates, the license prevalence rates, and case clearance rates for '97.
Each nonviolent (property) crime per 100 people was found to predict an increase of 53 licenses (s=12) per 100,000 population, and each one percent increase of minority population was found to predict a decrease of 7.6 licenses (s=1.9) per 100,000 population. The violent crime rate, median income and population density were not found to have significant predictive value. The nonviolent crime rate accounted for 21 percent of the license prevalence variation, and the minority portion accounted for 8.5 percent of the variation, for a total of 27 percent accounted for after adjusting for four extreme counties.
When the minority portion of the population was excluded from the regression, the portion of the population "in poverty," was found to have not quite as good predictive power, accounting for 19 percent of the variation while the nonviolent crime accounted for only 8.2 percent of the variation.
Each ten-fold increase in population density was found to predict an increase of 1.4 licenses revoked per 1000 licensees (s=.4). Population density accounted for 15 percent of the variation of revocation rates. None of the other factors included in the analysis was found to be significant for predicting revocation rates.
Because the data was available, a regression was run for the violent crime rate against the portion "in poverty," the median income, the minority portion of the population, and the population density. Minority population was found to be highly predictive. When the minority population was excluded from the regression, a combination of median income, population density and portion "in poverty" was found to be moderately effective, with all the coefficients being positive. Results were decidedly worse if either the income or poverty measures were dropped.
"Violent" crimes included murder, forcible rape, robbery, and "aggravated" (serious) assault. The "nonviolent" crime rates included burglary, larceny and motor vehicle theft.
License prevalence for each county was calculated by dividing the license numbers from the AG web site by the populations from the census bureau. Portions of the public "in poverty" was calculated for each county by dividing the number from the census bureau by the population.
Multiple linear regressions were performed for each of the two dependent variables (license prevalence and license revocation rate) as well as for violent crime rate.
PREVALENCE (/100K) = 498 + 53 x NONVIOL CRYM RATE (/100) - 7.6 x MINORITY%
The standard deviations for the factors were 63, 12 and 1.9, respectively. The regression "F" value was 13.43 for p<.001.
The median income, violent crime rate, and population density were not found to have significant predictive ability. The nonviolent crime rate accounted for 21 percent of the license prevalence variation, and the minority portion accounted for 8.5 percent of the variation, for a total of 27 percent accounted for after adjusting for four extreme counties.
When minority portion was excluded from the regression, the best regression was:
PREVALENCE(/100K) = 688 + 32 x NONVIOL CRIME RATE (/100) - 16 x POVERTY%
The standard deviations for the factors were 102, 12 and 4.5, respectively, and the "F" value was 11.9 for p<.001.
RATE (/1000) = -1.41 + 1.40 x LOG10POP'N/MILE2
The standard deviations for the factors were .83 and .39. The regression "F" value was 12.65 for p=.001. Population density accounted for 15 percent of the variation of revocation rates. None of the other factors included in the analysis was found to be significant for predicting revocation rates.
It is most likely that the explanation for the result is that the causality is actually in the reverse directionthat is, that the increased prevalence of concealed carry licensees causes increased rates of nonviolent (property) crime rather than the other way around. It has been found that states passing RTC concealed weapon carry laws causes reductions in violent crime and increases in nonviolent crime.[ 1 ] The logically satisfying explanation proposed for that result is that professional criminals deterred from robbery by the fact that people are likely to be armed turn to burglary and automobile theft, crimes less likely to bring the criminal in direct contact with the potentially armed victims.
If the existence of a RTC law would cause an increase in nonviolent crime, it is logical that greater numbers of people obtaining the licenses/permits would cause greater crime rate increasesbecause of criminals becoming aware of the prevalence of licensees or of increased incidence of their prey being armed.
With nonviolent crime rate eliminated as a possible cause of variations in concealed weapon license/permit prevalence, only the minority population is left as a potential cause (out of those factors considered in this study).
There has been no exodus of minorities from places where concealed carry becomes common, so the negative correlation between the prevalence rate and the minority rate cannot be from the former impacting the latter. The negative correlation between minority portion and concealed carry license/permit prevalence is probably for the most part because of the well established fact that minorities tend strongly to possess firearms less than whites (primarily because of lower income and multiple cultural factors).
The positive correlation of license revocation rate and county population density is very likely a result of law enforcement in metropolitan areas being "less friendly" to concealed carry licensees and licensing. Metropolitan law enforcement officials have typically resisted the passage of the right-to-carry laws. The fact that minority population portion was not found significant indicates that there was no perceptible racial discrimination in the revocation of licenses (although it cannot be ruled out in connection with the original issuance of licenses).
The fact that the best regression for violent crime rates included both the median income and portion "in poverty," and the fact that the coefficients for both were positive, suggests that income inequality plays a significant role in violent crime. This could result both from the fact that the inequality has an antisocial effect upon potential criminals and from the fact that criminality resulting in part from poverty is encouraged when there are relatively wealthy victims who will have the most things of greatest value.
1. Lott, John More Guns, Less CrimeUnderstanding Crime and Gun Control Laws, 2nd ed. Chicago, IL, U. of Chicago Press, 2000 p52
© 2001. Permission is granted by the author to reproduce this report in its entirety, or to reproduce any part of it with attribution.