-M W fANY MEMBERS of the press and the public apparently believe that interest groups buy influence by contributing to political campaigns. This belief has been cited as reason for Congress either publicly to finance congressional campaigns or to place a limit on the amount of monies that a candidate may receive from political action committees. In spite of this public discussion, few scholars have investigated systematically whether campaign contributions directly affect public policy or, more specifically, whether the likelihood that a legislator will vote for a bill favored by an interest group increases upon receiving a contribution from the group. Empirical research on legislative voting can be categorized into two groups. The first group of studies has investigated the relationship between dimensions of voting (e.g. composites of votes on several issues) and measures of general district opinion such as referendum results (Miller and Stokes, 1963; Erickson, 1978; Kuklinski, 1977, 1978; Kuklinski and McCrone, 1980; and Markus, 1974). The second group has focused on the relationship between a congressman's vote on specific bills and the opinions (or self-interest) of sections of his constituency (Bernstein and Anthony, 1974; Danielson and Rubin, 1977; Kau and Rubin, 1978, 1979; and Abrams, 1977). Opinion is usually measured by socioeconomic characteristics. To elaborate on Kuklinski's (1979) criticism, the presumption is that legislative voting is influenced by constituents' opinions but socioeconomic characteristics are no more than proxy variables for such opinions. Most of those used are poor proxies because a priori it is not clear which of many characteristics will predict voting and because the ability of a characteristic to predict voting presumably varies over issues. To be an adequate proxy, a socioeconomic characteristic must be sufficiently refined so that it is related to opinion on a specific bill. In the case of a vote on a specific piece of legislation, the size of a group for whom the issue is salient may be a convincing measure of its influence; and the more precisely the group is defined, the better the variable as a proxy. For instance, in this study dairy industry characteristics arguably are better measures of opinion on dairy price supports than general district characteristics such as percentage of the population which is rural. Furthermore, dairy industry characteristics presumably are poor measures of the relevant dimensions of district opinion on issues such as civil rights. For issues involving specific subgroups of votes, the relationship between a proxy variable and voting is not necessarily a black box. However, district level measures of