Analysis Explanation: To analyze the sequencing of democratic erosion, I use within-country fixed effects regressions (FEOLS), which allow me to compare each country to itself over time while controlling for persistent differences across countries. Because the original DEED dataset is organized at the level of individual events, I first restructure the data into a country–month panel. This involves creating a panel base that includes every country and every month between its first and last observation, including months in which no events occur. This produces a balanced panel with 42561 observations. Within this panel, I generate indicator variables for whether a given country-month contains a precursor event or a symptom event, and I construct a key explanatory variable that captures whether a precursor occurred within a specified time window (e.g., the previous 6 or 12 months), excluding the current month.
Using this structure, I estimate a regression for every possible precursor–symptom pair to determine whether specific precursor events tend to precede specific symptoms within countries over time. Each regression evaluates whether the presence of a precursor in the recent past increases the likelihood of observing a given symptom. I then focus on statistically significant relationships and summarize the strongest patterns. The estimated coefficients can be interpreted as changes in probability: for example, a coefficient of 0.06 implies that the occurrence of a given precursor is associated with a 6 percentage point increase in the probability of observing the corresponding symptom in subsequent months, holding constant country-level factors. This approach provides a systematic and interpretable way to identify which types of democratic erosion events tend to occur first.