Research on policy change tends to focus on legislative successes (bills that are enacted), policies that are especially important or controversial, and the final stages of the policy process. This article attempts to show how to improve our ability to trace support for policy proposals through the entire legislative process, for failures as well as successes and for less-visible proposals as well as more visible ones. We refine the concept of a "policy proposal"-a particular proposed solution to a public problem-as a set of identical or nearly identical bills introduced into one or more congresses; show how to find such bills, and examine a stratified random sample of 60 considered by the U.S. Congress; describe how much support the proposals receive; show that, in line with some views of legislative activity, proposals are generally on the agenda for only a short time; and suggest that trends in sponsorship provide a good way to measure support for particular proposals for policy change. It is argued that the approach developed in the article will aid subsequent studies of the determinants of policy change.
We have learned much in recent years about the causes of congressional action on policy proposals-about how Congress is affected by public opinion, the party balance, interest groups, and other factors. Our conclusions have been based, however, on a very partial picture of the policy process. Most research focuses on legislative successes (bills enacted, money appropriated, etc.), the final stages of the policy process, and important or controversial issues. We know relatively little about congressional action on bills that do not make it to the final stages of the legislative process, or are neither especially controversial or important-that is, the vast majority of bills (on these points, see, e.g., Baumgartner and Leech 1998: 38, 40; Talbert and Potoski 2002: 865; Edwards, Barrett, and Peake 1997: 547; Krutz 2000).
Why so partial a picture? We focus here on what we believe are two key reasons. First, we cannot explain the ebb and flow of congressional support for policy proposals from the time they are introduced because we cannot measure it. second, we can say little about congressional action on the vast majority of bills because the concept of random sampling has not been applied to the study of congressional action on policy-we have focused on biased samples of bills or policies without much concern for how this undermines our ability to generalize.
This article shows how to begin filling in our picture of the legislative process. We propose a way to measure congressional support for policy proposals, based on sponsorships and cosponsorships, and use the measure to gauge support for a (stratified) random sample of policy proposals from the 101st congress, 1989-1990.
Our measure of support is intended to be used as a dependent variable in subsequent studies of policy change, making it possible to test our theories of policy change over the whole policy process, studying the determinants of sponsorship as well as roll-call voting and enactment. And our approach to sampling is intended to improve our ability to generalize about congressional action.
LEGISLATIVE SUPPORT FOR POLICY PROPOSALS
Analyses of the determinants of legislative action most often try to explain enactment or some clear result of enactment, such as expenditures. Researchers know they should examine earlier phases of the legislative process, but seldom move back beyond roll-call outcomes (particularly in quantitative work).1 We thus know little about support for policy proposals never voted on, or about what distinguishes proposals that are voted on from those that are not.
There are two key reasons why we know little about support for such policy proposals. First, we have no satisfactory way to measure it. Second, when we try to develop such a measure, we realize that there is an even more fundamental problem: we have no good operational definition of the thing to be measured-of the concept of "policy proposal" itself. …