Wednesday, July 27, 2016

Per Capita Firearms vs Murder Rates in the U.S. 1950 - 2014

While various pundits and scholars have claimed that more guns equal more murders, the raw numbers reveal that to be untrue.  The graph above plots the per capita firearms rate in the United States (stated as number of firearms per 10 people) against the murder rate as tabulated in the FBI Uniform Crime Reports (UCR).  The UCR includes both murder and non-negligent homicide in the murder numbers.  The UCR murder rate is per 100,000 people in the United States.

The murder rate may be one of the most reliable numbers in the UCR, because murders are more likely to be reported than any other crime.  There is a body, and usually an investigation.  Other crime numbers are less certain. I have not found numbers for the UCR murder rate from 1945 to 1950, but a source, "Violence in America", by Gurr, states that the rate was fairly steady at about 5.8 per 100,000 in the 40s. The numbers could be measured from the next graph, but it would involve some measurement error. If the 45-49 numbers become available, I will use them in an updated chart.

Using the 65 years of data on the chart, the calculated correlation coefficient is a negative .1274, showing essentially no correlation.

The per capita firearms rate builds on work done by Newton and Zimring, then Gary Kleck. From 1950 to 1987, the data was taken from "Point Blank: Guns and Violence in America" by Gary Kleck, Table 2.1. 

The methodology used by Newton and Zimring, and then Kleck,was applied to the figures obtained from the ATF for later years.  The number shown is the cumulative addition of domestic manufacture plus imports minus exports.  This does not count guns shipped to the U.S. military.  Per capita numbers are calculated using the population figures from the U.S. Census.

There was as similar rise and fall in murder rates from about 1900 to 1958.  Initial rates are not as clear nationally in the early years.  Reporting was spotty before 1905.   5 murders per 100,000 range seems reasonable for the start. Then the rate rises to 9.7 in 1933.  The rate then starts the long decline to a minimum of 4.0 in 1958.

During that period, the per capita firearms were likely well below the .35 level calculated for 1945 by Newton and Zimring.  Those numbers are not available by year, so we cannot calculate a correlation coefficient.  If we could, it is clear that the results be similar to those from 1950 to 2014.  That is a large rise and a large fall in the murder rate, against a steadily rising per capita number of firearms.  Newton and Zimring calculated an increase of 47 million firearms added to the private stock from 1900 to 1945.

This reinforces the finding that the per capita number of firearms has little to do with the murder rate.

However, the number of firearms routinely carried for self defense might have an inverse relationship with the murder rate.

One intriguing note about the fall of the homicide rate from 1933 to 1958 is the high percentage of justifiable homicides that occurred at about the same time as the peak of 1933.

I am aware of two studies that separated out homicide motives during this period.  Those were noted in Violence in America, by Gurr.  The first study was done on Detroit data, by Boudouris.  Gurr writes that both studies showed high percentages of justified homicides, but does not state the numbers.

Boudouris found that Detroit from 1926 - 1934 had the highest homicide rate in the country. He found that the largest percentage of those homicides were non-criminal or justifiable homicides by police officers or private citizens.

The second study was on homicides in Chicago.  It found similar results.   From Violence in America:

An analysis of 883 homicides in Chicago for the years 1926-1927 showed a similar pattern. Like Detroit, a large percentage of homicides during this period were justifiable.
Shootings in 1926-1933 may have been more deadly, due to lack of antibiotics and advanced surgery techniques. 

These studies were independent of FBI statistics and did not use the extremely restricted and arcane Uniform Crime Report definition of a justified homicide.

The recent rise in justified homicides along with the increase in concealed carry permits and legal doctrines such as "stand your ground" and "castle doctrine" has been noted in several places. These reforms coincide well with the drop in the murder rate from 1991 to 2014.  Shall issue concealed carry reform started in 1987, and has continued through the entire 1991 - 2014 period, with the bulk of the reforms occurring after 1993.  This is the same period with the sustained drop in the murder rate.

The FBI UCR only catches about 10 to 20 percent of justified homicides, as confirmed by several studies. 

The temporal match of the drop in murder rate with a spottily reported rise in justifiable homicides coincides well with John Lott's thesis of "More Guns, Less Crime".

While intriguing, the poor reporting and recording of justifiable homicide numbers makes statistical analysis difficult.

©2016 by Dean Weingarten: Permission to share is granted when this notice is included.
Link to Gun Watch 

No comments: