We examine factors that influence whether or not constituents know and like their representative in Congress, exploring how constituents respond to very different forms of political representation-descriptive, symbolic, and allocational-provided by House members. We compare the relative contribution that non-policy representational factors make to the electoral advantage that incumbents enjoy among their constituents. The statistical analysis demonstrates that several non-policy aspects of the representational relationship, including descriptive correspondence and the member's legislative activity, benefit the incumbent through increased name recognition and, ultimately in voters' choices at the voting booth.
In the vast body of research that students of Congress have produced over the last several decades, the scholarly conversation over the nature of constituent representation stands as one of the longest-running and most interesting lines of inquiry. The classic work of Miller and Stokes (1963) sparked an exchange over the extent of issue congruence between representatives and the represented,1 but a broader discussion also emerged about the very definition of representation (e.g., Mansbridge 1998; Pitkin 1967; Prewett and Eulau 1969; Weissberg 1978; Williams 1998). Even though the issue of political representation remains of interest to political theorists, interest among empirical researchers during the 1970s began to shift somewhat from political representation to congressional elections (see Mezey 1993). Electoral politics, and members' ongoing pursuit of reelection, though, also pointed empirical congressional scholars toward broader conceptions of representational connections (Mayhew 1974).
The existing body of empirical work on political representation recognizes at least six different perspectives on the representation mechanism. In addition to the Miller-Stokes perspective on policy representation, two types of non-issue representation-service and allocation representationinvolve a focus on constituent and district benefits. A fourth perspective recognizes symbolic actions as part of the constituent-legislator relationship (Eulau and Karps 1977; Sinclair 1997), while others recognize a different sort of symbolism in the descriptive nature of representation (Canon 1999; Lublin 1997; Mansbridge 1999; Tate 2003; Whitby 1997). Still other work has taken an extremely broad view and described representation as a collective, national-level phenomenon (Mansbridge 1998; Page and Shapiro 1983; Weissberg 1978).
In this article, we ask how constituents respond to several of these non-policy aspects of the representational relationship+ or -and, in turn, how and when incumbents benefit electorally from these facets of representation. We test how member choices about activity (both symbolic and allocational) strengthen or weaken the electoral connection (Mayhew 1974), and we compare voter responses to these conscious member choices with constituent responses to the descriptive similarity between representative and voter. In addition, we examine whether legislative activity in Congress has any effect on whether constituents know and like their representative. In short, we offer new empirical evidence that the electoral relationship between member and constituent is affected both by what members do and by who members are.
REELECTION, MEMBER ACTIVITY, AND DESCRIPTIVE CONGRUENCE
Constituent Responses to Incumbent Activity
Through name recognition and positive constituent evaluations, House members enjoy stratospheric reelection rates. And, from Mayhew (1974), we have reason to believe that these advantages follow in part from members' own choices about their representational activity. Members engage in national policymaking as well as other decidedly non-policy activities (such as constituency service) in order to cultivate constituency support (Cain, Ferejohn, and Fiorina 1987; …