![Sample our Law journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5942/TOC-Subject-Sample-2021-Law.jpg"})
Abstract
Hate crime inflicts a variety of harms on victims, communities, as well as society at large. Scholars have long sought to understand the motivations and conditions behind hate crime offending. Green and his colleagues conducted the classic neighborhood studies examining the conditions that foster hate crime. Using data on hate crime in New York City from 1995 to 2010 from the New York Police Department’s Hate Crimes Task Force, the current study replicates and extends Green and colleagues’ neighborhood studies, investigating whether their findings hold true over an extended period of time in New York City as the city underwent major demographic changes. Using a group conflict framework, the current study extends prior work testing hypotheses derived from defended neighborhoods, social disorganization, and strain theories to explain ethnoracial hate crime.
Keywords:
Acknowledgements
I would like to thank Joshua Freilich for his invaluable guidance and support throughout this project, as well as Jeremy Porter, Amy Adamczyk, Jack McDevitt, and Peter Simi for reviewing earlier drafts of this article when it was part of my doctoral dissertation. I also thank Lieutenant Tara Coffey of the New York Police Department’s Office of Management Analysis & Planning, as well as Christopher Fisher and Rebecca Neusteter, for handling all of my data needs and questions.
Disclosure statement
No potential conflict of interest was report by the author.
Notes
1 While the current study does not control for time period, the inclusion of time-specific predictors controls for structural variations that may have occurred over time. Additional models were run with dummy variables for the three time periods and these analyses showed fairly similar results across all of the models; however, the current study presents the results without controlling for time as the significance of results is somewhat stronger in a number of models (i.e. p≤.05 versus p≤.1); although controlling for time leads to residential instability completely loses significance (p>.1) in a number of models.
2 All independent variables listed as coming from the US Census Bureau were downloaded from the IPUMS National Historical Geographic Information System (NHGIS) (Minnesota Population Center, Citation1990; Citation2000; Citation2012) via https://www.nhgis.org.
3 All 1995 variables computed as 1995=(1990 + 2000)/2.
4 Calculating the 5-year change periods necessitated the use of the Census Bureau’s American Community Survey for 2008–2012 as an indicator for 2010 population levels to calculate a 2005 midpoint for the appropriate change variable. The change variables for the three time periods are thus calculated as 1995–1990, 2000–1995, and 2005–2000, respectively.
5 The component excludes single female-headed households. Given current times, it seems inappropriate to consider female-headed households with no children as a marker of disadvantage as it seems more indicative of success than disadvantage (as it indicates the ability to own/rent a residence in the expensive housing market of New York City without the help of a spouse or a roommate).
6 The US Census Bureau’s (2001) formula is as follows: “the diversity index reports the percentage of times two randomly selected people would differ by race/ethnicity. Working with percents expressed as ratios (e.g. 63%=0.63), the index is calculated in three steps: A. Square the percent for each group, B. Sum the squares, and C. Subtract the sum from 1.00. Eight groups were used for the index: 1. White, not Hispanic; 2. Black or African American; 3. American Indian and Alaska Native (AIAN); 4. Asian; 5. Native Hawaiian and Other Pacific Islander (NHOPI); 6. Two or more races, not Hispanic; 7. Some other race, not Hispanic; and 8. Hispanic or Latino. People indicating Hispanic origin who also indicated Black, AIAN, Asian, or NHOPI were counted only in their race group (0.5% of the population). They were not included in the Hispanic group.” While the US Census Bureau’s (2001) formula uses eight different ethnoracial categories, the current study needed to use six categories as the 1990 US Census did not yet include separate categories for Asian and Pacific Islander variable as well as two or more races and some other race variable.
7 Worsening economic conditions require change variables just as used with demographic change variables. Thus, the 2000–2005 change variable is calculated using unemployment and poverty data from the Census Bureau’s American Community Survey 2008–2012, so a 2005 midpoint can be calculated.
8 Poisson regressions returned significant Pearson and Deviance goodness-of-fit statistics for all models, indicating negative binomial regression is the appropriate method. The likelihood-ratio test of alpha for the negative binomial regressions were significant for all models confirming overdispersion in the dependent variables and further demonstrating that negative binomial regression is the appropriate technique.
9 All models presented control for the unlogged Total Population, but are not included in the tables.
10 Additional analyses did examine anti-white hate crime in the context of Hispanic and nonwhite defended neighborhoods, but these models failed to show defended neighborhoods as a significant predictor of anti-white hate crime.
Additional information
Colleen E. Mills, PhD, is an Assistant Professor of Criminal Justice at Pennsylvania State University, Abington and she is a Project Manager for the US Extremist Crime Database (ECDB). Her research focuses on hate crime, far-right extremism and terrorism, racism, and group conflict. Her work has appeared in Crime & Delinquency, Race & Class, Homicide Studies, Studies in Conflict & Terrorism, and the American Journal of Criminal Justice.