Skip to main content

Do All Parties Die “The Same”? Using Hierarchical Competing Risks Models for Cross-National Research on Party Mortality

By: Published: 2019 | Product: SAGE Research Methods Cases Part 2
+- LessMore information
Search form
No results
Not Found
Download Case PDF


This case study highlights the advantages and challenges of using hierarchical competing risks models to analyze the determinants of party mortality from a comparative perspective. I review how these models can be used to simultaneously examine the impact of electoral, political, and institutional factors on two distinct but potentially correlated forms of party death, dissolution, and merger, while controlling for other observed and unobserved characteristics of the parties and of the democracies in which these operate. I illustrate the workings of this model by examining a data set covering the complete life cycles of 184 new parties that entered 21 consolidated democracies between 1968 and 2016. A key issue with hierarchical competing risks models is that standard statistical techniques and software packages for survival analysis either impose the assumption that the hazards (probabilities or risks) of both types of death are independent, or only model their dependence at the party or country level (but not both). Overcoming these limitations was the most important technical challenge faced during the project. In addition, over the course of the investigation, the members of the research team had to make several important methodological choices, such as how to select the parties to be included in the analysis, how to operationalize the different types of death, and how to deal with potential collinearity between the explanatory variables. I discuss how these challenges were handled in practice, and draw some lessons for researchers interested in party mortality and survival analysis more generally.

Looks like you do not have access to this content.

Methods Map

Cross-national research

Copy and paste the following HTML into your website