Contemporary representative democracies are facing multiple challenges. Among the most
important is the seemingly growing gap between political elites (party leaders, MPs) and
citizens. Public discourse and the media tend to consider that political elites have become
increasingly detached from the wishes and everyday realities of citizens. Subsequently,
established parties have drifted apart from their voters resulting in a lack of proper
representation of the citizens’ policy preferences, and thus contributed to the raise of populist
parties and movements in most of western democracies.
In contrast to public and media discourse however, political science research has delivered
mixed results in this regard. A number of studies show that political representation defined as
linking voters’ policy preferences to those of candidates and MPs and finally to enacted
governmental policies works rather well (e.g., Dalton, Farrell and McAllister 2013). On the
other hand, we also find studies which draw a negative picture of the quality of
representational processes, where representation only works for specific voter groups (e.g.
rich voters) and result in a generally large and growing distance between candidates’/parties’
positions and those of the voters. Thomassen (2012) addresses these contradicting results by
arguing that the methods applied to measure representation deserve more attention. He
hypothesizes that studies comparing party-voter positions based on the classic left-right
dimension provide much better results compared to studies based on a number of specific
policy issues which report clearly lower levels of congruence.
Based on data from the Swiss national elections 2015 our paper will compare the policy
preferences of candidates and their voters on the level of positions on a “catch-all” left-right
dimension as well as on a large number of specific policy positions. Data from the Swiss VAA
smartvote provides policy preferences for over 3’200 candidates (response rate of 84%) on 75
issues. Two pooled representative voter surveys conducted by the Swiss Electoral Study
(N=16’410) and one additional survey among smartvote users (N=11’109) contain identical
issue questions as the smartvote questionnaire and therefore allow for a direct and detailed
comparison of issue congruence.
In addition to its main focus on how good representational processes work and to which
degree political elites are detached from their voters, our paper will also dig into more detailed
views of the representational process, such as whether voters get better represented by the
MPs they voted for compared to all candidates on average, and whether we find voter groups
and political issues which are better represented than others. Finally, the paper will also speak
to VAA providers facing choices regarding the number of dimensions to be used and the
inclusion of saliency for matching procedures.
