Party Positions in the EP -- p4
As a matter of empirical focus, measurements of European party group positions have
focused either on measuring the group positions of party groups in the EP directly, or on
attempting to measure these indirectly through national-level measurements of the policy
positions of member parties. Direct measures include analysis of roll-call votes (Hix et al.
2005; Noury 2002), Stokman and Thomson’s (2004) expert interviews of political preferences
on 66 Commission proposals, and a survey of MEPs (Thomassen et al 2004), or analysis of
European manifestos (Gabel and Hix 2004).
Indirect measures rely on estimating the policy positions of party groups in the EP
through direct measures of the positions of actors associated with EP party groups. After
measuring the policy positions of these associated actors, the EP group position is assumed to
be the average of their associated groups. For instance, the European Election surveys used by
Thomassen and Schmidt (1997) measure the policy positions of mass publics and European
Election candidates associated with EP party groups. Alternatively, expert surveys of national
party positions, or CMP estimates based on national party manifestos, might be used to
estimate EP party group positions. For reasons we highlight later in this paper, however, it
should not be assumed that EP policy positions are always determined by the central tendency
of their national party members or mass public positions. Indeed, we see this degree of
convergence as one of the more interesting research questions to subject to empirical testing.
In terms of methodological divide, of course, there are numerous ways to measure the
policy positions of political actors, including opinion surveys, expert surveys, expert
interviews, analysis of party manifestos, and multi-dimensional scaling from roll-call votes.
Each method has advantages and disadvantages. Roll-call votes, for example, provide
objective political actions from which parties can be inductively located on policy scales,
using statistical techniques of multi-dimensional scaling. Roll call votes may suffer selection
bias, however, since they may be called selectively depending on political outcome and only
for certain issues. In addition, the substantive interpretation of the policy scales which they
produce must be interpreted, and these are not always clear (see Hix et al 2005).
The approach we use here is that of expert surveys: systematic placements by political
experts of party groups on numerous pre-defined policy dimensions. Expert surveys have by