jldbc / gunsandcrime

A replication of Marvell and Moody's economics experiment measuring impact of gun ownership on crime rates, using percent suicides by gun, gun manufacturing, and survey data as proxies for gun ownership. Data set included.

Guns and Crime

##A replication of Marvell and Moody's experiment of impact of gun ownership on crime rates

Marvell and Moody's study takes gun ownership data and measures its effect on crime ownership from 1977 - 1998. I have replicated their data, adding rows for years 1999-2013, and run an updated study.

Attached is the data set in .csv and .dta format, my team's write-up, and the STATA .do file for replicating the commands used.

This data is collected from public sources, so feel free to use and improve upon it as you please.

###The data set includes:

####Crime Variables: murder: murders per million population

rape: rapes per million population

robbery: robberies per million population

burglary: burglaries per million population

assault: assaults per million population

majorcrime: total major crime per million population

[crime measurements all from FBI Uniform Crime Report]

####Gun Ownership Variables: pgs: pct. of suicides by gun (proxy for gun ownership)

hg: pct. households owning handguns according to General Social Survey (better data than Gallup poll)

phg: imputed handgun measurement. Uses linear model to predit years when GSS did not ask this question.

gun: pct. households owning guns of any kind according to GSS

pgun: imputed gun, measured similarly to phg.

domestic_gun: guns available for sale in the US (a proxy - gun production - gun exports)

gallup_owngun: pct. households saying they own 1+ gun(s) in gallup polls

gunsamm: Guns and Ammo magazine circulation

amrmms: American Rifleman magazine circulation

amhmms: American Hunter magazine circulation

[gun measurements from GSS, CDC Wonder, Bureau of Alcohol, Tobacco, Firearms and Explosives, Gallup, and Alliance for Audited Media]

note: I could only find magazine data from 1999-2013. Duggan (2001) uses these as a proxy for gun ownership in the US. I believe this is a weak proxy, but I included what I could in order to attempt to test it.

####Control Varbables: Year: range 1980-2013

Pop: population in thousands

ampct: percentage population african american

metpct: percent of population living in urban areas

unrate: averge unemployment by year

prison: scaled prison population

employ: pct. population on standard payroll employment (nonfarm)

military: employed military personnel

rpci: real per capita income

r1524: pct. population age 15-24

r2544: pct. population age 25-44

r4564: pct. population age 45-64