1968) by asking whether legislators might simply be working harder, spending more time on the job, getting to know their constituents better, doing a better job representing them, and connecting more with their constituents. Perhaps it is just that academics are following the principle of the drunkard's search (look where the numbers are, rather than where the questions are [ Kaplan 1964: 11]). There is, to be sure, a part of the discipline that focuses on the dynamics of advocating policies and electoral reactions (for example, Sundquist 1983, Carmines and Stimson 1989), but this perspective does not seem as dominant as that evolved from the Mayhew framework. To the extent that framework is relied upon, it is less likely that the connections between constituents and politicians will be investigated. They are simply not seen as central to what keeps politicians in office.
Much academic research has a limited impact on politics. The development of the argument that class divisions have declined have, however, had considerable impact. The argument that Democratic Party positions on cultural, social, and race issues drove away the working class and hurt the party's chances in presidential elections ( Edsall and Edsall 1991b; Radosh 1996; Lawrence 1997; Carmines and Layman 1997) has become a fundamental part of interpreting American politics. Much of the evidence about this declining division was accepted because other research provided a basis for concluding that this conclusion made sense. If anything, this review suggests how powerful the impact of academic research can be in creating inclinations to accept particular conclusions and stop examining the evidence.