“…Estimating the effects of alliances is difficult with historical data because states do not form alliances randomly. We therefore fielded a large-scale preregistered survey experiment (5,6,7) on more than 14,000 voters in 13 NATO countries (Czech Republic, Denmark, France, Germany, Greece, Hungary, Italy, Netherlands, Norway, Poland, Spain, the United Kingdom, and the United States), with our sample data reweighted to match key demographic margins from each country's population.…”