UNITAS provides probabilistic estimates of political regime types (democracy, semi-democracy, autocracy) and transitions for 23,787 country-years from 1800 to 2017. Rather than relying on a single democracy measure or arbitrary cutoffs, UNITAS integrates information from 39 commonly used regime indicators using a Hidden Markov Model framework. The resulting dataset offers both modal regime classifications and full probability distributions, allowing researchers to incorporate measurement uncertainty directly into their empirical analyses.
Key Features
- Combines institutional, behavioral, and perception-based indicators into a unified measure
- Provides probability estimates for regime types, enabling soft assignment via multiple imputation
- Identifies regime transitions probabilistically, locating the most likely timing of change
- Captures both abrupt shifts and gradual, multi-year transitions
- Handles missing data efficiently without listwise deletion
- Enables analysis of longer regime sequences (e.g., sustained transitions, failed democratization)
How UNITAS Works
UNITAS models regime types as a sequential process, where classification in each year incorporates information from observable indicators and the historical trajectory of the political system.
Illustrative Cases
The figures below show Thailand (1980–2000) and Argentina (1965–1985). Bottom panels display classifications from individual regime indicators; top panels show UNITAS probability estimates and identified transitions. Note how UNITAS synthesizes disagreement among indicators: Thailand's 1991 coup registers as increased uncertainty rather than a full regime change, while Argentina's 1976 coup shows clear consensus supporting an autocratic transition.
Omer F. Orsun, Muhammet A. Bas
Political Analysis