This directory contains the four datsets
used in the paper "The Effect of Education on Criminal Activity: 
Evidence from Prison Inmates, Arrests and Self-Reports" 
The data are described in more detail in the paper.


********************************************************
* Census data merged with compulsory schooling data
* (inmates.dta)
********************************************************
The variables used in the paper are

prison: a dummy equal 1 if the respondent is in prison (derived from the group quarter variable)
educ  : years of schooling
drop  : high-school drop-out dummy
yearat14 : year in which the respondent was 14 years old
birthpl  : state of birth (fips code)
ca9  : dummy equal 1 if compulsory schooling is equal 9
ca10 : dummy equal 1 if compulsory schooling is equal 10
ca11 : dummy equal 1 if compulsory schooling is 11 or more
age  : age
black: dummy for black
state: state of residence (fips codes)
year : census year



********************************************************
* FBI arrest data merged with compulsory schooling data
* (arrest.dta)
********************************************************
The variables used in the paper are

state: state of arrest (fips code)
year : year of arrest
off  : offense (FBI code)
rage  : age group 14=20-4; 15=25-9; ... ;21=55-59.
educm : average education
dropm : percent high school drop-out
crime  : log of arrest rate
blackm : percent black
obsm   : cell size
yearat14 : year in which the respondent was 14 years old
ca9  : dummy equal 1 if compulsory schooling is equal 9
ca10 : dummy equal 1 if compulsory schooling is equal 10
ca11 : dummy equal 1 if compulsory schooling is 11 or more



********************************************************
* NLSY data
* (nlsycrime.dta and nlsyjail.dta)
********************************************************
The following two data sets are used to create Table 12.
Note:  Data on local unemployment rates and state of residence have not been included in 
either data set due to confidentiality of the Geocode data.


1. nlsycrime.dta 

- Data used for self-reported crime regressions (1980 survey data only)
- Includes all black and white males from the cross-section sample, poor white oversample, 
  and black oversample
- Description of variables:

id        NLSY identification code
white     dummy variable = 1 if respondent is white (cross-section white and poor white oversample)
black     dummy variable = 1 if respondent is black (cross-section black and black oversample)
violcr    dummy variable = 1 if respondent committed a serious violent crime in past year (positive 
          number of times used force to obtain things or attacked someone with the idea of seriously 
          hurting or killing them)
drugcr    dummy variable = 1 if respondent reported selling drugs in past year
propcr    dummy variable = 1 if respondent reported committing a serious property crime in past year 
          (positive number of times shoplifting or stealing something worth $50 or more from 
          someone/somewhere other than a store)
anycr     dummy variable = 1 if violcr=1 or drugcr=1 or propcr=1
sampwt    sample weight
hgc       highest grade completed (as of 1980 survey)
hsgrad    dummy variable = 1 if hgc greater than or equal to 12
enrly     dummy variable = 1 if enrolled in school last year
age       age in months (as of 1980 survey)
afqt      AFQT (revised) percentile
hgm       highest grade completed by respondent's mother
hgf       highest grade completed by respondent's father
teenmom   dummy variable = 1 if mother was a teenager at respondent's birth
reg*      dummy variable = 1 if living in region (s = South, ne = Northeast, nc = North Central, w = West) 
smsa      dummy variable = 1 if living in a SMSA


- additional remarks: Observations with hgc < 6 are deleted.



2. nlsyjail.dta 

- Data used for incarceration crime regressions
- Includes all black and white males from the cross-section sample, poor white oversample, 
  and black oversample
- Description of variables:

id        NLSY identification code
white     dummy variable = 1 if respondent is white (cross-section white and poor white oversample)
black     dummy variable = 1 if respondent is black (cross-section black and black oversample)
ja22t28   dummy variable = 1 if respondent spent any time in jail from ages 22 to 28 (incarceration
          status was determined from: (a) whether the respondent was surveyed in prison and (b) whether
          the respondent reported that he was incarcerated if asked about why he was not looking for
          work during an unemployment spell)
hgc22     highest grade completed (as of age 22)
hsgrad22  dummy variable = 1 if hgc greater than or equal to 12
age*      dummy variables = 1 if respondent is age 15-23 in 1980
afqt      AFQT (revised) percentile
hgm       highest grade completed by respondent's mother
hgf       highest grade completed by respondent's father
teenmom   dummy variable = 1 if mother was a teenager at respondent's birth
mreg*     fraction of years from ages 22 to 28 living in region (s = South, ne = Northeast, 
          nc = North Central, w = West) 
msmsa     fraction of years from ages 22 to 28 living in a SMSA

- additional remarks: Observations with hgc22 < 6 are deleted.