Premium
Response Scales in Voting Advice Applications: Do Different Designs Produce Different Outcomes?
Author(s) -
Rosema Martin,
Louwerse Tom
Publication year - 2016
Publication title -
policy and internet
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.281
H-Index - 26
ISSN - 1944-2866
DOI - 10.1002/poi3.139
Subject(s) - voting , advice (programming) , scale (ratio) , computer science , preference , data science , politics , political science , statistics , mathematics , geography , cartography , law , programming language
Voting Advice Applications (VAAs) represent popular election campaign tools in many countries, enabling voters to discover which party or candidate provides the best match with their political preferences. This article examines the effects of design choices on these tools by focusing on the response scale that is used to measure the policy positions of parties and voters. We analyze the impact of scale length on the advice generated by these tools using user data from a VAA developed for the 2014 Dutch local elections. We transform the original 101‐point scale into several alternative scale formats and determine if this leads to a different voting recommendation. We also examine the suitability of alternative scales for creating spatial models, which are often employed by VAAs. We show that the response scale has a potentially profound impact on the resulting advice (with voters receiving different VAA outcomes depending on the scale length), except for voters with an extremist response style. The findings have practical implications for the design of these tools: outcomes should be presented as a preference list, rather than focusing on the “best match.”