The methodological developments are taking us beyond the state of the art, and there have been several detailed, substantive results on distinct differences between sequences in successful episodes of democratization as well as autocratization and different types of failed episodes published - and some still in the works as working paper in the V-Dem series and currently under review. These results are already being used by a wide policy/practitioners community.
The pathbreaking "A third wave of autocratization is here" has been downloaded over 115,000 times and is already cited over 850 times in the literature. Among other things, we demonstrate there that a third wave of autocratization is indeed unfolding. It mainly affects democracies with gradual setbacks under a legal façade. A few other articles look at details, e.g. "The Institutional Order of Liberalization" reveal a clear pattern of reform during liberalization episodes with strong similarities across outcomes, but also that reforms to the administration of elections tend to develop comparatively earlier in episodes that produce a democratic transition; "Episodes of liberalization in autocracies" providing a description and analysis of all 383 liberalization episodes from 1900 to 2019, offering new insights on democratic “waves”. We also demonstrate the value of this approach by showing that while several established covariates are valuable for predicting the ultimate outcomes, none explain the onset of a period of liberalization.
In the second subproject seeking to advance methods for causal inference using observational data, some breaking-through articles include "How to Make Causal Inferences with Time-Series Cross-Sectional Data under Selection on Observables" and "Generalized Nonlinear Difference-in-Difference-in-Differences": Difference-in-difference-in-differences (DiDiD) allow for the correction of unmeasured con-founding and function as a robustness check for difference-in-differences (DiD) techniques. Athey and Imbens (2006) provides a scale invariant, nonlinear DiD approach known as Changes-in-Changes (CiC). Sofer et al. (2016) extends CiC by showing that pre-treatment outcome measures are a special case of placebo (negative) outcomes and proposes a generalization of CiC called Negative Outcome Control (NOC). We develop a generalized nonlinear DiDiD approach we call NOCNOC that can be used either in the traditional DiDiD setting or when a placebo outcome is available in the pre and post-treatment data. We show that NOCNOC can correct for bias in Di-DiD, CiC, and NOC.
