Analyzing Data on Hate Crimes in America from 1991 to 2019

In 2021 hate crime remains a terrible reality — one that has been in the news a lot recently. As a first step towards making sense of the current situation, let’s look at what the data shows for this class of crimes.

Dataset

The most detailed, comprehensive, and uniform dataset on hate crimes that I have been able to find is complied and released annually by the FBI. The most recent dataset covers up to 2019. It includes fields such as the state in which the crime was committed, demographic information on the offender and victim, and the type of offense, ranging from pocket-picking to murder.

Trends

*All demographic terminology is the FBI’s, not my own.

The graph above shows the victimization rates for several groups going back nearly 30 years. This reveals a number of trends:

• For most groups, hate crime rates steadily decreased from the late 1990’s to the mid-2010's.
• Then from the mid-2010’s, rates began to rise again across the board.
• Over this entire time period, crimes rates against Black people, Jews, and gay men have been consistently 10 to 20 times higher than for other large demographic groups.
• Gay men have the highest victimization rate of all members of the LGBT group, though anti-transgender crime is quickly approaching similarly high levels.
• Crimes against American Indians and Alaskan Natives surged in 2017.
• Anti-Islamic crime had been relatively low until spiking massively in 2001. Since then, it never returned to its previous levels and spiked again in 2017.

That last point came as a shock. The data truly suggests that Islamophobic crime after 9/11 was not just a temporary wave, but a decades-long downturn for the safety of Muslims in America.

Here are victimization rates per one million group members for every group:

178.91, Jewish
168.00, Sikh
156.39, Gay (Male)
142.14, Transgender
77.97,  Islamic (Muslim)
75.44,  Arab
63.09,  Black or African American
50.27,  Native Hawaiian or Other Pacific Islander
35.39,  American Indian or Alaska Native
29.66,  Lesbian (Female)
26.50,  Eastern Orthodox (Russian, Greek, Other)
23.18,  Multiple Races, Group
23.08,  Other Race/Ethnicity/Ancestry
22.09,  Lesbian, Gay, Bisexual, or Transgender (Mixed Group)
13.06,  Asian
12.53,  Hispanic or Latino
6.15,   Jehovah's Witness
5.33,   Buddhist
5.25,   Bisexual
4.51,   White
4.48,   Hindu
3.81,   Mental Disability
2.73,   Mormon
1.98,   Physical Disability
1.59,   Catholic
0.48,   Female
0.30,   Atheism/Agnosticism
0.24,   Protestant
0.15,   Male
0.07,   Heterosexual

The 2020 Question

The FBI data for 2020 will likely not be available until around November 2021. But in the meantime, various organizations have reported a spike in crimes against Asians.

The Center for the Study of Hate and Extremism at California State University, San Bernardino reported a 145% surge in Anti-Asian crime from 2019 to 2020.

Their data only covers a selection of major cities, but if we assume a similar trend nationwide, that would bring the FBI total from 253 in 2019 to 620 in 2020. This raises the rate from 13.06 to 32.00, ignoring any population growth. Adding this data point to the chart shows a victimization rate at levels not seen for this group since the early 2000's.

Time will tell what the official data shows for 2020 and onward, but hopefully this will not follow the same pattern as anti-Muslim crime, in which rates never fully subside after an initial spike.

Methodology

First, the total number of victims per group is calculated for each year. This is not the same as the number of incidents, since a single incident can have multiple victims.

The next step is to convert these raw numbers into per capita numbers. Without this step, it is difficult to interpret the data. For example, the data from 2019 lists a total of 880 Anti-White and 154 Anti-Arab hate crime victims. Of course, every single crime affects a real person and is worth calling out. But to understand the overall trends, we should bear in mind that the non-Arab White population is nearly 100 larger than the Arab population in America, meaning that the rates per capita are actually over 16 times higher for an Arab individual.

It was not easy to find good sources of data to cover each of these many categories, e.g. what was the population of bisexual people in America in 1993 vs. 1994? I cannot promise that the population numbers I used are perfect, but they should be sufficient for the analysis here, where victimization rates often differ by orders of magnitude. See the github link for more details on how population figures were calculated, along with the full data, more graphs, and the code used.

Additional Graphs

The following graphs contain all available demographic categories from this data, including groups not shown in the charts above. Note that some categories were not counted separately until the 2010's.

Open Questions

Here are some questions and hypotheses that this dataset alone does not address. Perhaps other researchers can provide answers/evidence for or against them.

• What has driven meaningful decreases in hate crime in the past and will these approaches work for the current situation?
• Does society, media, and policymakers tend to react more to a group with a changing rate than one with a stable high rate? Is there an element of forgetfulness, fatigue, or “writing-off” the situation when it is consistently high?
• Are the rates for Anti-Hispanic crime in this dataset an accurate representation? If a subset of this group does not report a hate crime to the authorities for fear of deportation, this figure could be underreported. What other groups may be more or less willing or able to report a crime?
• In general, what could be some flaws in the the FBI dataset or this methodology? What other data sources or analysis could help clarify the picture?