DATA 712 Extra Credit

Hate Crimes in NYC: Modeling the Impact of National Events (2003–2023)

Introduction

This project investigates how hate crimes in New York City changed from 2003 to 2023, especially around the time of major national or international events. The focus is on four groups that have historically been targeted: Jewish, Asian, Black, and Muslim. Some of the key events studied include the October 7th Hamas-Israel war (2023), the COVID-19 pandemic, the murder of George Floyd (2020), the Tops supermarket shooting in Buffalo (2022), and the Muslim Ban (2017). The data for this analysis comes from the FBI’s Uniform Crime Reporting (UCR) program, which maintains hate crime data from 1991 onward. However, I limited the scope of this study to the years 2003 through 2023, as this period ensures consistent annual reporting across all four targeted groups. I hope to see if each of the events highlighted are statistically significant linking it to the increases in hate crimes during that spike.

Research Method

Data Source

The dataset used in this study comes from the FBI’s Uniform Crime Reporting (UCR) Program, which records hate crime incidents reported by law enforcement agencies across the United States. The analysis focuses specifically on incidents reported in New York City from 2003 to 2023. Each incident includes details about the bias motivation, offender characteristics, and location.

Statistical Analysis

Given that the outcome variable (annual count of hate crime incidents) is a count variable and overdispersion was expected, both Poisson and Negative Binomial regression models were fit for each group. Dummy variables were added for years corresponding to major social events hypothesized to influence hate crime frequency.

Final model selection was based on Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). All models were implemented in R and the results were formatted using the modelsummary package for clean presentation.

Results

To conduct the analysis, I applied both Poisson and negative binomial regression models—methods used for analyzing count data, such as the annual number of reported hate crimes. While the Poisson model is a standard starting point, because the hate crime data showed clear signs of overdispersion (the variance was much larger than the mean), the negative binomial model provided a better fit.

By creating event dummy variables to flag years in which major incidents took place, I was able to see which events held significance. For example, I marked 2023 for anti-Jewish hate crimes due to the Israel-Gaza conflict and 2018–2021 for anti-Asian hate crimes related to the COVID-19 pandemic. These dummy variables allowed me to estimate whether those years were associated with statistically significant increases in hate crimes for the affected group, while also examining longer-term time trends.

The results revealed clear patterns. In 2023, during the Israel-Gaza conflict, there was a statistically significant increase in anti-Jewish hate crimes. The model showed a notable spike for that year (p = 0.008), suggesting a strong likelihood that the increase was event-related rather than random. A more modest, borderline-significant increase occurred in 2019 (p = 0.056), which aligned with a series of high-profile attacks against Jewish communities across the United States. These findings suggest that both international conflicts and domestic hate crimes elsewhere can trigger a rise in local incidents, even when New York City is not directly involved.

In the case of anti-Asian hate crimes, the model indicated a steady and statistically significant upward trend over time (p = 0.002). The COVID-19 pandemic years were associated with visibly elevated counts, though the specific event dummy variable covering 2019–2021 did not reach statistical significance. This suggests that anti-Asian hate might have been increasing prior to the pandemic and that COVID-19 likely intensified this existing trend rather than initiating it.

For anti-Black hate crimes, the analysis showed a statistically significant decline over the 20-year period (p < 0.001). However, 2022—the year of the Tops supermarket mass shooting in Buffalo—stood out as an exception. That year showed a statistically significant increase in anti-Black hate crimes (p = 0.045), possibly reflecting a reactive surge tied to the national attention surrounding the racially motivated attack. Other highly visible events, such as the murder of George Floyd in 2020 and the death of Eric Garner in 2014, did not result in statistically significant changes in NYC’s reported incidents, which may reflect local variation in responses or broader community resilience.

Anti-Muslim hate crimes remained relatively stable across the two decades, showing neither a consistent upward nor downward trend. However, the year 2010—the height of the Ground Zero mosque controversy—was associated with a statistically significant increase (p = 0.041). Other moments, such as the 2017 Muslim Ban, did not produce statistically significant changes in the data, though they may have shaped public discourse in less measurable ways.

## [1] "Model comparison for: Anti-Jewish"
##               Model      AIC      BIC
## 1           Poisson 602.9755 611.9546
## 2 Negative Binomial 355.5937 366.0693
## [1] "Model comparison for: Anti-Asian"
##               Model      AIC      BIC
## 1           Poisson 361.8881 365.1613
## 2 Negative Binomial 155.0206 159.3848
## [1] "Model comparison for: Anti-Black"
##               Model      AIC      BIC
## 1           Poisson 404.7100 413.6890
## 2 Negative Binomial 295.7221 306.1976
## [1] "Model comparison for: Anti-Muslim"
##               Model      AIC      BIC
## 1           Poisson 166.8018 172.0244
## 2 Negative Binomial 147.4391 153.7062
Negative Binomial Models of Hate Crimes in NYC (2003–2023)
Anti-Jewish Anti-Asian Anti-Black Anti-Muslim
+ p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001
(Intercept) 15.396 -224.545** 96.113*** -106.835**
(0.181) (0.003) (<0.001) (0.005)
Year -0.005 0.113** -0.046*** 0.054**
(0.367) (0.002) (<0.001) (0.004)
Pittsburgh_2018 0.305
(0.291)
Kosher_supermarket_Rabbi_House_Shul_shootings_Poway_shooting 0.551+
(0.056)
Operation_pillar 0.144
(0.484)
Israel_Gaza_War_2023 0.782**
(0.008)
COVID_2019_2020_2021 0.910
(0.155)
Trayvon_2012 -0.118
(0.765)
Floyd_Taylor_2020 0.391
(0.331)
Graner_2014 -0.303
(0.458)
Tops 0.797*
(0.045)
GZ_Mosque_2010 1.650*
(0.041)
MuslimBan_2017 -0.144
(0.780)
Park51 -0.962
(0.151)
Num.Obs. 33 22 33 21
AIC 355.6 155.0 295.7 147.4
BIC 366.1 159.4 306.2 153.7
Log.Lik. -170.797 -73.510 -140.861 -67.720
RMSE 40.48 23.22 18.98 6.68

Conclusion

This analysis demonstrates that hate crime patterns in New York City are not entirely random but often shaped by national and international events. Local spikes in hate crime reports appear to follow moments of intense public discourse, political conflict, or violence involving specific groups. These patterns were most pronounced for Jewish and Muslim communities, where geopolitical events and domestic controversies were followed by measurable increases in hate incidents. For Asian and Black communities, both long-term social trends and violent events shaped hate crime dynamics in distinct ways.

Understanding these connections can help improve prevention efforts. When specific types of events are known to influence hate crime activity, city agencies, schools, and community organizations can prepare accordingly. While events like these aren’t predictable, responding promptly when they do occur allows support systems to be mobilized, community outreach to be expanded, and proactive measures to be taken before tensions escalate.