Generating High-Quality Verifiable Relational Data about News Coverage of Protests and Other Events from Electronic Archives

Pamela Oliver, University of Wisconsin
Alex Hanna, Google
Chaeyoon Lim, University of Wisconsin

We demonstrate the value and importance of modifying standard practices in the collection of protest event data from electronic archives. Both the physical events involving real people and actions and the portrayal of those events in news sources matter. Events have relations to each other and are described in relation to other events. Better ways of collecting and organizing data can capture these relations while working faster and more efficiently. We build on our own past work in developing MPEDS as a tool for locating particles about protest and propose a strategy for a hybrid machine-human workflow. We recommend: (1) generating and releasing event-report data files or an event-report crosswalk file linking event records and report records; (2) changing the standard workflow in hybrid machine-human coding to preserve the original textual material about events in event-report files and defer coding into higher-level closed-ended categories or decisions about threshold or inclusion criteria to the pre-analysis phase, not the original data-collection phase, where machines can aid in batch-coding variables; (3) recognizing and coding relations among events using concepts of campaigns, episodes, and master- and sub-event relations; (4) analyzing events and reports about events in relation to each other.

No extended abstract or paper available

 Presented in Session 234. Empirical and Theoretical Advances in Protest Event Analysis