🔎 Research Focus

Examines how post-treatment bias skews efforts to separate economic and cultural drivers of the globalization backlash and whether common "horse-race" regressions can reliably apportion their relative influence on voting.

🧭 Key Methodological Warnings

• If—and insofar as—cultural variables are post-treatment with respect to economic factors, including them in regressions biases estimates of the effect of economic shocks on voting (and the reverse holds when economic variables are post-treatment).
• For the same reason, horse-race regressions do not permit accurate estimation of the relative roles of economic versus cultural factors.
• Changes in a factor's regression coefficient after adding post-treatment controls cannot be taken as evidence of mediation.

🧪 Evidence and Approach

• Replicates and expands earlier studies of the globalization backlash to evaluate the presence and magnitude of post-treatment bias in common empirical strategies.
• Presents novel cross-country analyses that probe the culture–economy nexus and illustrate how empirical inferences shift when post-treatment dynamics are present.
• Uses these empirical exercises to show how including post-treatment controls alters conclusions about voting responses to economic shocks.

📌 Why It Matters

Results imply caution in interpreting regression comparisons that include variables potentially downstream of key causal factors. Studies that use horse-race regressions to decide whether economic or cultural explanations drive the globalization backlash—or that infer mediation from coefficient changes after adding controls—may draw misleading conclusions unless the temporal and causal ordering of variables is carefully addressed.