Bivariate model for associations between traumatic events and suicidal behavior1.

*Significant at the .05 level, two-sided test.1Assessed in Part II sample due to having Part II controls. Each row represents a separate bivariate model; some models do not include data from all countries if the country is missing the variable. India and Brazil were dropped in the bivariate model for Combat, Exposure to War and Refugees; and Brazil was dropped in the bivariate model for Natural Disaster. For Israel, the entire sample is coded “Yes” for exposure to war with the age of onset set to the age they moved to Israel. Controls for all models included person-year, country, demographic factors (age, sex, time-varying education, time-varying marriage), interactions between life course (3 dichotomous dummies representing early, middle, and later years in the person's life) and demographic variables, parent psychopathology, and childhood adversities (additional details available upon request).