looking beyond district boundaries, policy representation may be evaluated on a collective basis
(Weissberg 1978). Thus, while individual legislators’ may not accurately reflect constituent
preferences on a district-by-district basis, the legislature as a whole can exhibit
policy/ideological tendencies reflecting the nation.
Perhaps the oldest and most prominent debate regarding representation arises when
evaluating policy representation by individual legislators. Specifically, should legislators act as
“delegates” or as “trustees?” The former notion focuses on the correspondence on an issue-by-
issue basis between constituent opinion and representative behavior. In this view, legislators
maintain no independence, instead acting only with explicit guidance from their constituents.
Alternatively, representatives may be regarded as agents of their constituencies that have been
entrusted with interpreting the constituents’ interests (which may or may not be the same as
those preferences actually expressed by constituents at any point in time) and acting on their
behalf with considerable discretion.
At the same time that disagreement about the proper definition of representation endures,
much research puts aside definitional questions to empirically assess the extent to which
legislators represent constituents given a certain conceptualization of representation. Much of
this empirical work has its roots in Miller and Stokes’ (1963) analysis of policy agreement
between members of Congress and their constituents across three issue domains. Using
correlations between constituent and legislator preferences, Miller and Stokes find that members
do seem are responsive to their constituents (although this linkage varies in strength by issue
domain). While many authors have questioned Miller and Stokes’ methodology and conclusions
(e.g. Achen 1977, 1978; Clausen 1973; Erikson 1978; Fiorina 1974), there is a consensus that
constituents exert at least some influence on their representatives. Consequently, many
2
Convention