Zayne Sember

I am a Marketing Data Scientist on the Model Maintenance Team at Recast, a marketing measurement company, where I help maintain and improve the models that power its clients’ marketing analytics.

Before Recast, I used computational methods to study social systems—in particular, how elections are administered in the United States. From 2025 to 2026 I was the Research Director at the MIT Election Data and Science Lab (MEDSL), whose mission is to improve the democratic experience for all U.S. voters. Managing a team of research staff and students, I oversaw MEDSL’s election datasets, led novel academic research, and helped produce flagship data products like the Elections Performance Index.

My own research asks when and how members of Congress respond to the people they represent. I treat legislators’ public communications—press releases, speeches, social media—as text-as-data: my dissertation, Timely Talk, traces how members adapt what they say as their districts and the national mood shift, and related work looks at election administration and affective polarization.

Outside of work I enjoy building data science projects. Most of my development time now goes to pressR, an R-based data collection toolkit and growing dataset of nearly every press release issued by a member of Congress—436,000+ releases and counting, back to 2010. I also created speccurvieR, an R package for fast, pretty, and easy specification curve analysis—version 1.0, now on CRAN, grows it into a full workflow for estimating, testing, decomposing, and reporting the curve.