When it comes to anticipating or determining the impact of global events, who is left out of or underrepresented in typical sentiment datasets and what can we learn when we hear from them? In this September 2022 Harvard University’s Center for Geographic Analysis talk, RIWI’s Danielle Goldfarb explains one way we hear from those typically under reflected in sentiment data, and what we can learn from them. Danielle discusses the design principles, advantages, limitations, and applications of Random Domain Intercept Technology (RDIT). RDIT is an “asking” technology that leverages the Web’s architecture to collect randomized and anonymous spatiotemporal data on people’s opinions and behaviors. The approach was originally developed to address some of the biases associated with conventional surveys. The randomized approach results in the inclusion of those who are typically excluded or underrepresented in most sentiment datasets. Data collection is continuous and time-stamped, and contains latitude and longitude information.
Danielle’s talk discusses a wide range of current academic and policy applications. These include measuring perceptions of military conflict escalation in Russia/Ukraine and China/Taiwan, Chinese sentiment towards the US, misinformation and disinformation, election prediction, economic monitoring, Russia-China relations, and real-time Ukrainian migration intentions.
View the recording here