Measuring policy congruence

IN MOST STUDIES OF POLITICAL REPRESENTATION, THE PRIMARY TOUCH-stone for ‘good representation’ is that elected politicians act in accordance with the preferences of their electorate. In cross-sectional studies this criterion is usually called ‘policy (or issue- or ideological-) congruence’ and the main debate is whether majoritarian electoral systems or proportional representation (PR) systems produce higher congruence (e.g. Huber and Powell 1994; Miller et al. 1999; Blais and Bodet 2006; Powell 2009). In longitudinal studies it is often called ‘policy responsiveness’, with studies debating whether some representative institutions adapt to changes in public opinion more quickly (Stimson et al. 1995), and whether responsiveness is caused by representatives adapting to voters, or the other way around (Esaiasson and Holmberg 1996; Holmberg 1997). All such studies agree that policy congruence (or responsiveness) indicates good political representation. This communis opinio, however, hides considerable disagreement about the proper operationalization and measurement of policy congruence/responsiveness.

Ideally, in order to gauge policy congruence, we need to compare the policy preferences of voters with the policy preferences of representatives. Voter preferences are customarily measured in surveys, often using a general Left-Right scale or specific issue scales. Occasionally, voter preferences are not measured directly, but are considered to be revealed by their party preference: all voters for a party are assumed to agree with that party’s manifesto. Obviously, such a strategy biases the results in the direction of high congruence. The preferences of representatives are sometimes measured through content analysis of election manifestos, unrealistically assuming that all representatives belonging to a party agree with all proposals in that party’s manifesto. As most of such studies in the manifesto approach count words or sentences devoted to particular policy areas, they measure a party’s issue saliency rather than a party’s issue position. A lively debate has ensued over the question whether policy positions can be derived from saliency measures (see e.g. Laver 2001). Alternatively, some studies employ expert surveys to measure the policy positions of the various political parties (also assuming that there is no variation among representatives of a party).