Hate Crimes in the U.S.
Recently, we’ve seen a significant rise in hate crimes in the U.S., and anti-Asian violence in particular has jumped to headlines in the last few months. Despite the swelling tide now, hate crimes aren’t new to the U.S., and the long history of prejudice-driven crimes highlights the difficult and ongoing work that is addressing hate in America.
Thus, I decided to analyze hate crime data with the goal of understanding biases and better informing government and organization leaders to combat hate crimes.
The questions that motivated my analysis:
- How has the quantity and bias type of hate crimes changed over the years?
- Where do most hate crimes take place?
- What is the relationship, if any, between an offender’s race and the bias type motivation of the crime?
The Data
The data was collected from Kaggle on hate crimes in the United States from 1991 to 2018. The complete dataset can be found here and contains over 200,000 records of hate crimes. The original data comes from FBI: Crime Data Explorer. For each report, there is information about the date, location, victim and offender counts, offender race, and bias type of the crime.
Hate Crimes Over Time
First, a general look at the total number of hate crimes recorded in the dataset by time.
We see that overall, the quantity of hate crimes spiked around 2001 and has been declining in recent years, with a bit of another tick up after 2015.
Then, I was curious to see whether there is a seasonality to hate crimes, or certain months where officers can expect more reports.
Not unexpectedly, there does appear to be some seasonality, with the number of hate crimes declining in winter months (Oct-Jan) and peaking just before in September. This can be explained by the weather being too cold for people to be out and about, leading to less hate crimes, or simply that the “good vibes” of the holiday season reduces such crimes.
Location
Next, I looked at how hate crimes are distributed by location.
Hate crimes are largely concentrated in states with major cities and more diverse populations, like California, New York, and Michigan. However, we should note that these states also have greater overall population, so a direct comparison can’t easily be made without standardizing to account for differences in population.
Interestingly, while we often hear about hate crimes committed by random street people or at public locations, most hate crimes actually occur at residential areas or at home. This suggests that many hate crimes are extremely targeted and premeditated; in other words, offenders intentionally seek out the residences of victims. Another explanation is that prejudice exists even among similar communities.
Bias Types
I also examined how hate crimes have affected individual demographic categories over time by grouping hate crimes by bias types. Since the original dataset includes multiple bias-type offenses (eg. “Anti-Female;Anti-Mental Disability;Anti-White”) where there may only be 1 or 2 reports, I filtered out the top bias types for more informative visualization.
We see that Anti-Black or African American are the most common hate crimes types, and the trend of the Anti-Black category largely follows that of the total hate crimes over the years. However, comparing the two plots, while there is the spike in the total quantity of hate crimes in 2001, Anti-Black crimes actually decreased during that time. Moreover, none of the top 5 bias types appear to experience this peak. Thus, I plotted the top 9 (without the Anti-Black category to fit better on the same scale) bias types to see which groups drove the increase in hate crimes after 2000.
The plot shows that ultimately, the Anti-Other Race/Ethnicity/Ancestry category experienced a sharp rise in hate crimes in 2001. This may be attributed to post-9/11 unrest and xenophobia.
Also notable is the lag of the Anti-Gay, Anti-Lesbian, and Mixed Group categories; while these groups also follow a trend where there is a rise in hate crimes followed by a decline, the peak occurs later for these groups than for the racial bias types. This shows that while the US may have become more tolerant of racial diversity, acceptance of different sexualities has been a more delayed and drawn out process.
Offender Race and Victim Demographic
Lastly, for each bias type category, I plotted the percentage of hate crimes committed by an offender of a specific race.
We see that most hate crimes are committed by an offender of unknown or white race. In particular, most Anti-Black hate crimes are committed by an a White offender, and most Anti-White hate crimes are committed by a Black or African American offender.
Conclusion & Next Steps
Ultimately, this analysis provides insight into the progression of hate crimes in the US over time by bias and location motivations, but there are certainly limitations and next steps to consider.
It’s important to note that hate crimes are generally underreported, with marginalized communities being less likely to report. Moreover, the motivation for this study was largely driven by recent events, where we’ve seen an increase in racially-driven violence and media coverage. However, this dataset only includes hate crimes up until 2018, limiting my analysis to older reports. A natural next step would be to factor in more recent data once it’s available and analyze the extent to which the quantity and motivations for hate crimes have actually changed in light of recent events.
Additionally, it would be interesting to combine this analysis with other datasets on state population demographics and government policies. For example, we could look at the correlations between a state’s percentage of population in poverty or in higher education and the number of hate crimes. We can also examine whether certain state laws and strength of law enforcement plays a role in the makeup of hate crimes.
____________________________
About the author
My name is Olivia, and I’m a junior at the University of Pennsylvania Wharton School. This data project was conducted in R for the course OIDD245: Analytics & The Digital Economy.
