In my article, “Separation-of-Power Games in the Positive Theory of Congress and Courts” (Segal 1997) I compared the theoretical and empirical validity of the separation-of-powers model and the judicial attitudinal model as applied to U.S. Supreme Court decision making. I found that the preferences of the justices often fell inside the set of irreversible decisions, and that even when the preferences fell outside the set, the justices, with the exception of Earl Warren, showed little evidence of constraint. I concluded, consistent with the attitudinal model, that the justices overwhelmingly engaged in rationally sincere behavior.

Tim Groseclose and Sara Schiavoni (1998), whom I thank for their efforts, have alerted me to two errors in my calculation of the set of irreversible decisions. First, I used the formula 2CC-G (where CC = the conference committee and G = the gatekeeper) for calculating indifference points in the constraint set rather than 2G-CC. Second, I implicitly allowed Congress to pass legislation that would be vetoed and could not be overridden. Correcting this error changes the optimal behavior of the conference committee, which will not propose legislation that can be vetoed but cannot be overriden, and the gatekeepers, who must consider the conference committee's actions.

In this note I replicate the study using the corrected data and find first that the sets of irreversible decisions are a bit broader than I had originally found. This means that fewer justices could be constrained by Congress than I had originally concluded.