Yesterday I noted Rachel Barkow’s terrific paper, which makes a strong case that originalist methodology influenced the non-capital jury votes of Justices Scalia and Thomas, leading them to support criminal defendants in circumstances that they otherwise might not.
I would like to return to that paper today for a different reason. Subscriber to the lawcourts and conlawprof listservs will know that legal scholars frequently post requests wanting to know, say, the percent of time that Scalia and Thomas voted together in criminal rights cases, or the percent of times that O’Connor or Kennedy joined 5-4 decisions, or how often the court issued judgments rather than opinions during the Warren, Burger, and Rehnquist Courts, etc. Rachel’s paper was a bit more complex, trying to examine the voting behavior of Scalia, Thomas, Rehnquist, Kennedy, and O’Connor in capital and noncapital criminal procedure cases when they did not vote as a bloc.
One way to do this, as Rachel did, was by hand, combing through the U.S. Reports or one’s favorite online source. An alternative is to use Harold Spaeth’s U.S. Supreme Court database. I just fired up STATA and in less than five minutes I knew that Scalia and Thomas have voted together 92.0% of the time in criminal procedure cases; that through their careers O’Connor joined 68.43% of the Court’s 5-4 decisions and Kennedy joined 69.2% (these figures could readily be broken down by term, natural court, issue area, etc.), and that the number of cases resulting in judgments of the Court under Warren, Burger, and Rehnquist was 69 (1.93%), 149 (2.77%), and 93 (2.59%) respectively. Rachel’s problem is more complex and would probably take a little more time, but the point is still the same.
There are tradeoffs to using Spaeth’s database, of course. Judgment calls have to be made, and in using someone else’s data you need to rely on that person’s judgment. I do note, though, that unlike many datasets, Spaeth provides intercoder reliability tests.
There are also start-up costs: learning to use an appropriate statistical software program (the database comes in SPSS, SAS, STATA and ascii formats); and learning to use the database appropriately. E.g., the database can be configured to include or exclude secondary issues and laws included in the case, additional dockets from the same case, cases with or without oral argument, decrees, tied votes, etc. I do note that Lee Epstein and I plan to provide a workshop next year on using the Spaeth database, and a few other well-used ones too.
For now, ELS scholars can find databases to their hearts delight housed at the University of Kentucky’s Ulmer Project