In just 13 pages, Warren Miller and Donald Stokes's “Constituency Influence in Congress” established the agenda for the next half-century of research on congressional representation. Along with its sister paper, “Party Government and the Salience of Congress” (Stokes and Miller 1962), “Constituency Influence” famously documents the general impoverishment of voters' knowledge of Congress and its members. Of at least equal importance is its analysis of congressional behavior as a response to oft-ambiguous constituency views, incorporating rich interview data of the sort not seen since. It is no wonder that Miller and Stokes' paper is among the most heavily cited—and influential—of all papers that have graced the pages of the Review.

At the heart of “Constituency Influence” is the well-known “diamond model” of political representation. Consider a\break diamond with constituency opinion at the left node and roll-call voting at the right node. The path from constituency to roll call is moderated by the representative's personal position (top node) and perception of constituency opinion (bottom node). One pathway from constituency opinion to roll call voting is via electoral selection: constituency opinion→representative's opinion→roll-call behavior. The second pathway is via the representative's anticipation of electoral sanctions: constituency opinion→representative's perception→roll-call behavior. Miller and Stokes added the nuance of reciprocal causation, whereby representatives' personal views color their perceptions of constituency opinion while the perceptions of constituency opinion modify their personal views.

For over 100 U.S. House districts across three issue domains, Miller and Stokes estimated constituency opinion via district-level aggregates from the 1958 Survey Research Center (later NES) postelection surveys and applied standard techniques to measure the representatives' roll-call positions. The truly innovative aspect was the measurement of the personal opinions of the representatives themselves as well as their perceptions of constituency opinion via the Congress members' responses in personal interviews. No study before or since has been so bold as to connect members' personal views to those of their constituents and their roll-call voting in this way.

“Constituency Influence” must be judged in the context of its time. Like virtually everything set to print by the four authors of The American Voter (Campbell et al. 1960), “Constituency Influence” displays a gracefulness to its prose that appears remarkable when read today. It introduced the methodological repertoire of “causal modeling” and “path analysis” to the pages of the Review, albeit with much of the detail embedded in the footnotes. Like any ambitious paper, “Constituency Influence in Congress” has some limitations. Typical for empirical papers of its day, the statistical presentation is sketchy, preventing easy replication. The various measures are on different metrics, so that one cannot say, for instance, whether representatives are to the left or right of their districts or whether their public postures are to the left or right of their private beliefs. (This, of course, is a limitation that frustrates contemporary work on representation as well.) The major limitation certainly was the estimation of constituency opinion with sample sizes that current standards hold to be too small for reliable analysis.

“Constituency Influence” concluded with a seemingly out-of-equilibrium result: “Congressmen feel that their individual legislative actions may have considerable impact on the electorate, yet some simple facts about the Representative's salience to his constituents imply that this could hardly be true” (p. 54). While noting the obvious possibility that members of Congress have an inflated sense of their visibility to constituents and perhaps could get away with more than they think, Miller and Stokes to their credit also saw the possibility that members of Congress know what they are doing. Miller and Stokes set the questions that remain central today regarding how much and by what mechanisms a not very political public achieves representation by its elected leaders.