cover image: Forecasting Electoral Violence - David Randahl, Maxine Leis, Tim Gåsste, Hanne Fjelde, Håvard Hegre, Staffan I. Lindberg,

Forecasting Electoral Violence - David Randahl, Maxine Leis, Tim Gåsste, Hanne Fjelde, Håvard Hegre, Staffan I. Lindberg,

1 Nov 2024

Whereas the holding of elections in authoritarian and hybrid regimes has aggravated the challenge, the problem is not restricted to the Global South or to weakly consolidated democracies, as recent episodes of electoral violence in the United States, Brazil, Turkey, and Hungary remind us. [...] The first indicator is the Election Government Intimidation (v2elintim), measuring the extent to which the government uses intimidation and harassment to influence the outcome of elections. [...] Our indicator for the level of electoral violence is calculated by collapsing the original scale of the two indicators into three categories: no electoral violence, for elections where the indicator has the value 3 or higher; moderate electoral violence, for elections where the indicator is in the range 1.5-3, and severe electoral violence for elections where the indicator is smaller than 1.5. [...] The final ensemble weights were then used to combine the predictions of the constituent models to make the final forecasts for 2024 and 2025 so that the predicted probability of each election violence level is the weighted average of each constituent model’s predicted probability. [...] The prediction of electoral violence specifically is a challenging task due to the complexity of the phenomenon and the multitude of factors that can influence its occurrence (Birch, Daxecker, and Höglund, 2020; Höglund, 2009; Birch and Muchlinski, 2018).

Related Organizations

Pages
31
Published in
Sweden

Table of Contents