Response Scales in Voting Advice Applications: Do Different Designs Produce Different Outcomes?