In the aftermath of the Capitol riots that took place on January 6, 2021 in Washington, D.C., federal prosecutors continue to charge participants involved. USA TODAY is gathering details of these cases as the FBI keeps identifying and prosecuting more perpetrators involved in the storming. They have compiled a list [Link] of people arrested on a range of charges filed by federal prosecutors, as well as those arrested by Capitol Police and D.C. Metro Police for storming the capitol building.
Information for each arrest includes the date of arrest/charge, home state, the charges and a brief summary of what happened. We have collected data from the first arrest taking place on January 6th up until February 24th accounting for a total of 244 arrests. Taking forward our analysis done on Capitol Riots, we have drawn inferences from these arrests using different visualisations and highlighted some interesting findings.
In order to gain a better understanding of what the “Charges” and “What Happened” fields were discussing, we did content analysis by creating word clouds for the two fields. Using them, we were able to observe key words of the types of charges that were more prevalent against the Capitol rioters, their involvement in the event and how they were eventually arrested.
In the word cloud for the Charges field, we observe that the majority of the words present in it such as demonstrating, picketing, abetting, violent, destruction, assaulting, obstruction, parading, disruptive, unlawful, disorderly, conspiracy that depict the nature of the activities in which the protestors indulged. Furthermore, the presence of words that have violent connotations such as weapon, violence, violent, destruction, assaulting, deadly point towards the violent nature of some of the activities which the protestors were charged with.
We noticed that in the word cloud for What Happened field the most prominent words are Capitol, FBI, court which is expected as this field discusses how the different Capitol rioters were identified and arrested by the FBI. This is further supported by the presence of the words such as arrested, arrest, reported, identified, witness, tip, criminal. An interesting observation was the presence of words related to different social media platforms such as social, media, posted, photos, footage, video, Facebook, Instagram, account. This indicates that there was a significant use of social media platforms as well as the photos and videos that are posted on them in the FBI investigation of the riot.
Additionally, we created a sankey plot to represent what were the different charges and how they are distributed. In order to create this plot, we manually annotated all the charges into thirteen broad categories. Some of these categories were further divided into smaller categories if there was use of dangerous weapons. For instance: Unlawful Entry (all) is divided into Unlawful Entry and Unlawful Entry with Dangerous Weapon. In case the charge specified who the crime was against, then we further divide the broad category for that charge – for example Obstruction (all) is divided into Obstruction of Law Enforcement and Obstruction of Justice. Using our annotations, we were able to identify 705 charges against a total of 242 people who were arrested. We observe that the most common charge is unlawful entry with 223 occurrences. At least 7 of these cases involved the rioter carrying a dangerous weapon. The next most frequently occurring charge is Disorderly Conduct with 214 such charges, out of which 25 were against an officer or employee of the US government with 2 of the cases involving dangerous weapons. There is also presence of crimes that are aggressive in nature such as Violence, Destruction of Property, Illegal possession of firearms, Theft and Threats. A detailed version of the Sankey plot can be found here.
We further wanted to explore the impact social media had on the arrests. We observed that the FBI used social media posts to a large extent to find and arrest rioters. As of February 24, 2020, 1:43 PM IST, 244 arrests were made in regard to Capitol Riots. Facebook was instrumental in identifying 99 (40.5%) of them, followed by Twitter, contributing in 63 (25.81%) of them. We also saw a number of other social media platforms involved in the arrests, with a total of 10 unique social media platforms. The arrested people had posted their photos and videos on these platforms or were identified in those posted by others. The individual counts for each platform can be found in the graph below:
We also wanted to inspect on how different states were participating in the riots. On conducting a frequency based analysis of states and the arrests made , we see the results reported below :
- A total of 237 arrests were made with the state mentioned in them. A total of 42 states are there where arrests have been reported.
- With arrests in a state peaking at 24 in Pennsylvania, followed by Florida, Texas, New York respectively. Top 6 states alone accounted for close 50% of the 237, showing a great deal of concentration of these arrests.
- We see a number of states where the number of states where arrests have been in smaller numbers.
The choropleth map above shows a coloured geographical distribution of the state wise arrests made. The states marked as S are Swing States in the elections. The list of swing states was taken from here.
Diving deeper , we see the following insights from the data :
|Arrests from Blue States||128|
|Arrests from Red States||109|
|Arrests from Swing States||90|
54% of the total arrests have been made in states where Democrats won the 2020 elections.
Around 38% of all the arrests made till February 12, 2021 have been from Swing states, out of which, 47.8% have been in Blue States. As visible from the map, the top three states with the highest number of arrests are Swing states.
The dataset used for our research can be found here. We would like to thank USA Today for making the the data public for researchers like us to analyze it.
Students involved in the analysis — Manvith Reddy, Arundhati Bhattacharya, Pankil Kalra, Samiya Caur, Tushar Mohan, Vibhu Agarwal, Vishwesh Kumar.