New research uncovers a surprising pattern in measuring support for violent extremism: over time, responses tend to shift toward alignment with local political power due to the Grapevine Effect. This phenomenon occurs as participants hear about the study and alter their answers out of fear of retribution.

Drawing from original data collected across three Sahel countries through surveys on sensitive topics, we document this pattern clearly. We systematically ruled out alternative explanations—such as changes in public opinion or interviewer effects—and demonstrate that even experimental methods are vulnerable to these response shifts.

This finding has profound implications for political science research: it cautions against conventional survey approaches and highlights the need for more nuanced data collection strategies when measuring sensitive attitudes anywhere.