When Surveys Get Interrupted: How Unexpected Events Reveal Causal Effects

🧾 What This Paper Covers

An increasing number of studies use unforeseen events that occur during survey fieldwork to estimate causal effects. This identification strategy—termed the Unexpected Event during Survey Design—relies on sudden, salient events that effectively split respondents into treatment and control groups. The paper examines this approach in detail, spelling out the assumptions needed for valid inference and highlighting the main threats to identification.

🔍 How the Strategy Works and What’s Tested

• The strategy treats the timing of an unexpected event during fieldwork as a source of quasi-random variation between respondents interviewed before and after the event.
• Special attention is given to the observable and testable implications of the assumptions that justify causal claims.
• Potential threats to identification are catalogued and tied to concrete checks that researchers can perform.

📊 Evidence and Illustration

• Uses data from the European Social Survey (ESS) to show how the method can be implemented in practice.
• Illustrates the discussion with an original empirical application: the impact of the Charlie Hebdo terrorist attacks (Paris, 01/07/2015) on French citizens’ satisfaction with their national government.

✅ Recommendations for Researchers

• Proposes a series of best practices framed as estimation strategies and robustness checks that lend credibility to causal estimates obtained from unexpected events during surveys.
• Emphasizes the importance of testing observable implications and of transparent reporting of identification assumptions.

⚠️ Why It Matters

The Unexpected Event during Survey Design is a promising and increasingly popular tool for causal inference in public opinion research, but its validity depends on clear assumptions and careful diagnostics. This paper provides practical guidance for scholars who wish to exploit survey interruptions while avoiding common pitfalls.