Assessing the applicability of vote advice applications for estimating party positions

We have recently witnessed a considerable interest in the estimation of party preferences from vote advice applications (VAAs). Such attempts have been met with some skepticism. One major potential concern is the effect of the item selection on the resulting spaces in scaling analyses. The items are said to introduce an implicit weighting, such that the estimated spaces may or may not be indicative of party competition. We assess the severity of this issue by employing explicit weighting schemes based on external sources. Three such sources for calibrating VAAs are proposed and subsequently tested using data from the 2017 GermanWahl-O-Mat. We show that the position estimates are remarkably stable. Moreover, the item count in ordinary VAAs is sufficiently large to ensure robust position estimates.