Forum metadata on views and responses enables quantitative study of hate group engagement patterns

Evidence

Stormfront is ideal for the current research purposes for several reasons. First, its members pride themselves on their intellectualism, and they prohibit any discussion of violence on the website. Second, each discussion thread is presented as a separate entity, which can be read or responded to independent of other discussion threads. Additionally, each thread is listed under the subject header that was used by the person who initiated the thread. Finally, next to each subject header is a pair of numbers indicating how many people have viewed the message and how many have responded to it. Thus, it is possible to quantify the interest level generated by each topic, at the level of both viewing and responding. We took advantage of these features to assess the types of messages that were read and responded to most frequently, and whether people were more likely to read and write in-group-enhancing messages or out-group-derogating messages. — Roderick M. Kramer, Geoffrey J. Leonardelli, Robert W. Livingston, Social Cognition, Social Identity, and…, 988

Categorization

• SEED: The methodology itself is an angle: using natural online community metadata (views, replies, thread structure) as unobtrusive behavioral data to study extreme groups without survey reactivity.
• QUESTION: View-count vs. response-count distinction could reveal silent consumption of hate content vs. active engagement — a behavioral attention metric worth exploring further.