The purpose of the project was to learn more
about patterns of homicide in the United States by strengthening the
ability to make imputations for Supplementary Homicide Report (SHR)
data with missing values. Several strategies for imputing missing
homicide data have been suggested in recent years. However, aside from
assessing their logic and the degree to which findings based on
imputed data are in accordance with what is known broadly about
homicide, such strategies are difficult to evaluate for the reason
that the "true" values for cases with missing data cannot be
observed. By combining different datasets available through the
National Archive of Criminal Justice Data, this project offered a
strategy for both testing imputation approaches that have been
suggested in the literature and for using these results to advance the
development of new methods for imputing SHR data.
In order to better understand why data are missing
and how missing and nonmissing data differ, Supplementary Homicide
Reports (SHR) and local police data from Chicago, Illinois, St. Louis,
Missouri, Philadelphia, Pennsylvania, and Phoenix, Arizona, for 1990 to
1995 were merged to create a master file. The researchers used
overlapping information on victim and incident characteristics from
the SHR and police data from fields for which the data are very rarely
missing even if the offender is unknown (age, sex, and race of
victims, month of incident, and weapon) to link the two types of data
sources. While this process became challenging when the
characteristics of victims and incidents were identical across
multiple incidents in a given month, this type of situation was
extremely rare as usually at least one characteristic varied across
incidents. Through this process, 96 percent of the cases in the SHR
were matched with cases in the police files. The data contain
variables for three types of cases: complete in SHR, missing offender
and incident information in SHR but known in police report, and
missing offender and incident information in both. The merged file
allows estimation of similarities and differences between the cases
with known offender characteristics in the SHR and those in the other
two categories. The accuracy of existing data imputation methods can
be assessed by comparing imputed values in an "incomplete" dataset
(the SHR), generated by the three imputation strategies discussed in
the literature, with the actual values in a known "complete" dataset
(combined SHR and police data), for both case-by-case imputation as
well as distributions at the aggregate level.
Not applicable.
Homicides known to the police in Chicago, Illinois,
Philadelphia, Pennsylvania, Phoenix, Arizona, St. Louis, Missouri,
from 1990 to 1995.
homicides
CHANGING PATTERNS IN SOCIAL POLICY IN PHILADELPHIA,
PHOENIX, AND ST. LOUIS, 1980-1994 (ICPSR 2729). Data for St. Louis
were supplied by the St. Louis Homicide Project (Scott Decker and
Richard Rosenfeld, principal investigators).
HOMICIDES IN CHICAGO, 1965-1995 (ICPSR 6399)
UNIFORM CRIME REPORTS [UNITED STATES]: SUPPLEMENTARY
HOMICIDE REPORTS, 1976-1999 (ICPSR 3180)
administrative records data
Variables from both the Supplemental Homicide
Reports and the police report offense data include incident date (year
and month), victim characteristics (age, sex, and race), offender
characteristics (age, sex, race, and relationship to victim), and
incident details (weapon used, circumstance, and situation of
offense). Data also contain geographic variables (UCR state code,
agency, population group, region, population, county, and SMSA), as
well as matching information, such as from which source the offender
data were missing, the number of victims in the SHR data, the number
of offenders in the SHR data, the researchers' confidence in the case
match, their comments, and several aggregated and derived variables.
Not applicable.
None