It is clear that a number of data sets will be necessary to test the nine propositions laid out in the previous chapter: roll-call votes, committee membership, issue areas, and so on. We set aside this chapter, though, to examine the one data set unique to the present study, namely, a measure of executive discretion. Our theory provides us with two starting points in this task. First, as our model centers on the passage of legislation, our proper unit of analysis is a law. Second, discretion in our model is composed of both delegation and constraints – the former giving power to the executive and the latter circumscribing its limits.

In constructing our discretion index, we are forced, to some extent, to rely on individual judgment. But to minimize subjectivity, we devised a set of systematic coding rules and procedures for each element of the data set. In this chapter and its accompanying appendixes, we describe our sample of laws, detail how we code and construct our key variable of interest, and present some summary statistics and trends in executive discretion over time.