Item

Abstract

We propose a novel approach for generating time-series for party positions as an alternative to the estimates provided by the Manifesto Project. Our approach combines multiple expert surveys from different years, filling up the missing data using a multiple imputation algorithm that uses additional information from mass surveys. We illustrate this approach by estimating time-series for eight European countries for periods up to 50 years and show that our estimates are comparable, if not superior, in richness and face validity to those of the Manifesto Project. The data can be used to explore the robustness of empirical analyses using party position data and serve as valid benchmarks for computational text scaling and crowd-sourced manual coding of party manifestos.