Thread by @austinlwright

We are releasing a new paper on the #insurrection event in Washington, DC

We use a new method for estimating community-level participation in mass-scale protests. We find a robust link w/ partisanship, Proud Boys and Parler posts.

Link: bit.ly/36w02bZ

@packlesshepherd

We validate a method for identifying protest sites using high frequency cellphone data that we wrote about here: bit.ly/2YvO0ef.

Leveraging device surges + media documentation of the `March to Save America' + geotagged Parler videos, we can pinpoint the event location.

After locating the protest site using the mobile device data, we can identify the community of origin for each device that likely participated in the event.

Historical records of origin-to-destination movement suggest the communities participating in the event deviated significantly from previous visits to the Washington area.

We then study the community-level factors associated with participation in the Jan 6 event in DC. We combine novel data on voting patterns at the census block group (CBG) level, measure proximity to the nearest HQ of the Proud Boys organization, and intensity of Parler uploads.

We find evidence that partisanship, socio-political isolation, proximity to chapters of the Proud Boys organization, and the local activity on Parler are associated with protest participation.

These findings are robust to a number of alternative model specifications.

We believe our research helps fill a prominent gap in the study of collective action: identifying and studying communities involved in mass-scale events that escalate into violent insurrection.

NB: Working with data of this type naturally raises important ethical questions, which we carefully considered. Our analysis uses only anonymous, aggregated data and at a geographical level that complies with the US Census Bureau’s own standards on disclosure avoidance.

This research is relevant to a story that broke yesterday by @jimrutenberg @jobeckerhrw @EricLiptonNYT @maggieNYT and co @nytimes

This work leverages data/archives compiled by @splcenter and @donk_enby.