I’m all for attempting to bring empirical analysis to the study of law, but unfortunately it’s very hard to do well and in a way that tells us anything meaningful. Take the New York Times’ recent attempt to identify a leftward shift in the Supreme Court based on a database that codes decisions as “liberal” or “conservative.” It sounds like the database is widely used and accepted as a standard, but that doesn’t make it particularly accurate.
Here’s quick indicator that the dataset is questionable: there are 9-0 decisions on both sides of the spectrum. So they’re coding, for a significant number of cases each term, Justice Ginsburg’s votes as “conservative” and Justice Thomas’s as “liberal.” If that’s the case, one has to question what they even mean by liberal and conservative.
And there’s the rub. Do the database’s editors mean decisions favoring parties who are themselves liberal or conservative? Decisions whose outcomes track a typically liberal or conservative political policy agenda? Decisions based on conservative or liberal judicial philosophies? And, more importantly, which of these indicators would tell us anything useful about how the judges are functioning?
My best guess about this database is that their coding is a very crude shorthand for political stereotypes about legal issues. Take one very broad swath of cases: decisions favoring a criminal defendant are apparently coded as liberal. This blurs countless lines. For example, Justice Scalia spearheaded a trend in taking the Confrontation Clause more seriously, thus throwing out large quantities of prosecutorial evidence. Is that “liberal” just because it helps defendants, or conservative because it takes the Constitutional text seriously? The database codes it as “liberal.”
Recent cases pushing back on overcriminalization – a common issue of concern among conservatives – also code as pro-defendant and therefore “liberal.” But the means by which the Court reaches a pro-defendant result can be liberal or conservative themselves. Take Bond v. United States, last year’s case addressing the use of a federal chemical weapons ban to prosecute a woman who used chemicals to harass her husband’s mistress. One hundred percent of the justices agreed that the defendant should win, but while three justices would have held (using conservative modes of analysis) that Congress didn’t have the constitutional authority to criminalize common domestic crime, the other six employed the more liberal method of interpreting the defendant’s conduct right out of the statute’s clear language. Yet the study codes the decision as a 9-0 “liberal” result.
Similar cases arise in other areas of law as well. Was the 9-0 result in McCullen v. Coakley liberal or conservative when it overturneda Massachusetts’ abortion clinic buffer zone law on First Amendment grounds? One might code it as a “conservative” decision because it helped pro-lifers and hurt the abortion industry. Others might say it was a liberal decision because it strengthened free speech (never mind that some of the strongest proponents of free speech on the Court are conservative). The database actually coded it as both, deciding that the (5-4) holding on content neutrality was liberal and the (9-0) holding on narrow tailoring was conservative. But how is it that a position on content neutrality that is actually more restrictive of free speech is “liberal,” while a strong narrow tailoring requirement that protects free speech is conservative?
This is not to say that there is an easy way to determine such things. It’s only to say that the task is so challenging that the simplistic coding in this database is virtually useless. It’s definitely possible the Roberts Court is lurching to the left. But I wouldn’t base my judgment on such a limited study.