Person or Place? A Contextual, Event-History Analysis of Homicide Victimization Risk, United States, 2004-2012 (ICPSR 37079)

Time Period(s) View help for Time Period(s)

Date of Collection View help for Date of Collection

Data Collection Notes View help for Data Collection Notes

Study Purpose View help for Study Purpose

The purpose of this research was to examine the influence of neighborhood social disorganization on the risk of homicide victimization focusing on how community effects changed once individual-level characteristics were considered. The researchers sought to understand how disadvantage functions at neighborhood- and individual-levels interact to influence homicide risk. This research integrated concepts from social disorganization theory, a neighborhood theory of criminal behavior, with concepts from lifestyle theory, an individual theory of criminal behavior, by examining the effects of both neighborhood-level predictors of disadvantage and individual attributes which may compel that person to behave in certain ways. This research examined nationally representative data from the 2004-2012 National Health Interview Survey (NHIS) linked with National Death Index data, which provided individual-level data on homicide mortality both achieved characteristics and ascribed characteristics. The 2005-2009 and 2008-2012 American Community Surveys (ACS) were integrated using geographic identifiers from the NHIS.

Research questions for this study include:

1. What are the effects of characteristics of the neighborhood on the risk of homicide victimization net of individual characteristics?
2. How do individual factors (such as age, sex, race, immigrant status, level of education, employment status, marital status, dependent children, military experience, health insurance, etc.) condition the effects of neighborhood-level factors (such as collective disadvantage, social structure, and race-specific urban composition) on the risk of homicide victimization?

The primary contribution of this research was to investigate whether the influence the neighborhood environment or a person's own demographic and social characteristics, along with the type of lifestyle that persons with those characteristics may lead, played a larger role in the risk that a person has of being a homicide victim. Is it the "kind of person", the "kind of place", or some combination of the two that has the largest influence on a a person's risk of being the victim of a homicide?

Study Design View help for Study Design

The study involved two distinct levels of secondary analysis because explanatory variables were measured using both individual- and neighborhood-level data. The individual-level units of analysis were individual respondents from the restricted-use 2004-2012 National Health Interview Survey (NHIS) with the National Death Index data (NHIS-MCD) linked file. The neighborhood-level units of analysis were block groups representing neighborhoods using data from the 2005-2009 and 2008-2012 American Community Survey (ACS).

Event-history, surveylogistic regression analyses were used to investigate the impact of neighborhood-level disadvantage on the risk of homicide mortality in block groups in the United States and examined the effect of individual-level characteristics on the association between context and the probability of an individual being the victim of a homicide. SAS software was used to analyze the data. Due to variation in the number of years each respondent was at risk of homicide victimization, a separate observational record (or person-year) was created based on the respondent's NHIS interview year and the number of years the respondent was at risk for homicide victimization.

Sample View help for Sample

The surveylogistic procedure in SAS allowed the sampling design of the National Health Interview Survey (NHIS) with National Death Index data (NHIS-MCD) to be considered, adjusting estimates based on sampling design information. Due to the complex NHIS sample design, strata and cluster (PSU) identifiers were included in order to specify which stratum or cluster each observations belonged. A weight variable was also included to adjust for potential bias due to variation in the number of individual observations associated with each respondent. The eligibility-adjusted weight measure included in the NHIS-MCD linked file was adjusted for the pooled structure of the file by dividing the NHIS weight measure by the number of years that were being pooled. Due to the redesign of these measures for the 2006-2009 survey years, the variance estimation measures (stratum and PSU) were converted for consistency across sampling designs. The data from the distinct sampling designs (2004-2005 and 2006-2009) were treated as statistically independent.

Time Method View help for Time Method

Universe View help for Universe

1. Individual respondents from the restricted-use 2004-2012 National Health Interview Survey (NHIS) Death Index-Multiple Causes of Death (NHIS-MCD) linked file.
2. Block groups representing neighborhoods using data from the 2005-2009 and 2008-2012 American Community Surveys (ACS).

Unit(s) of Observation View help for Unit(s) of Observation

Data Source View help for Data Source

Data Type(s) View help for Data Type(s)

Mode of Data Collection View help for Mode of Data Collection

Description of Variables View help for Description of Variables

This study included individual-level variables from the National Health Interview Survey (NHIS) with National Death Index data (NHIS-MCD) linked file, dealing with the demographic, geographic, family and household, citizenship and nativity characteristics of respondents. The NHIS-MCD provided individual-level data on homicide mortality both achieved characteristics (i.e. education, employment status, marital status, dependent children, military experience, health insurance, etc.) and ascribed characteristics (i.e. race, age, sex, immigrant status, etc.). At the neighborhood-level, primary indicators from the American Community Survey (ACS) that were examined included poverty, female headed households with children, educational attainment, public assistance, racial/ethnic heterogeneity, population aged 15 to 24, female headed households in poverty with children, grandparent households with children, marital status, citizenship, vacant housing units, home-ownership, and population 14 to 24 no job and not in school or military. Community-level characteristics of disadvantage, included measures of resource deprivation, urbanness, and housing instability that existed within each respondent's place of residence were obtained from the 2000 Census.

Response Rates View help for Response Rates

Not Applicable

Presence of Common Scales View help for Presence of Common Scales

None

Original Release Date View help for Original Release Date

2018-09-25

Weight View help for Weight

The National Health Interview Survey with National Death Index (NHIS-MCD) data included an eligibility adjusted weight measure.