If you can't please everyone,must you only please yourself?

Department of Government and Politics, University of Maryland, College Park, MD 20742 8221, U.S.A. Shirking models, especially those of Kalt and Zupan, have demonstrated that Senators often vote contrary to their constituents' ideology. These models establish two components of Senators' ideology through a regression of constituency demographics on interest group ratings. The predicted scores are constituency attitudes while the residuals are Senators' personal ideologies. Senators' personal ideology is presumed to be independent of constituency factors. The use of demographics is problematic, because it is unclear that they are good surrogates for attitudes. Using statewide estimates of ideology from public opinion surveys, I show that demographics provide reasonable estimates of public attitudes. However, estimates of shirking from public opinion depend upon constituency characteristics, a finding that is inconsistent with shirking models based upon residualization. The existing shirking models depend heavily upon a legislator's party as a key component of constituency opinion. But party is an attribute of the Senator and not of the electorate. A better interpretation is that Senators respond to their fellow partisans in the electorate.