Judging on the margins

by on

Sports are full of advanced statistics today: they are ways of trying to quantify the messy feelings we have when we watch them. How do we know Lionel Messi is so good? Because he’s constantly in the top right of the graphs. Why not apply the same approach to understanding judges and courts?

Using the Supreme Court Database, we can calculate a variety of advanced statistics. The analysts there have done all the hard work of breaking cases into individual issues, and recording how each judge voted on every issue.

So I wrote some code, and calculated out a couple of more advanced per-judge stats. The leaderboard follows after the explanation of the stats. I’ll try and keep it updated when the Database gets updated.

VC (Votes Cast): How many times a justice cast an issue on a vote.

MVC (Marginal Votes Cast): How many times a justice could have changed the outcome of an issue at the Supreme Court level if they had voted the other way. A justice gets one MVC for every time they form part of a one-justice majority, plurality, or a tied vote. This metric is, I think, key. It helps tell us which justices most directly influenced the path of Canadian law, even without necessarily writing judgments themselves. Before you scroll down, take a guess: which judge had the most impact? (I bet you’re going to be wrong!)

MVC/VC: A metric of marginal efficiency. Some justices aren’t around for so many issues, but for the time they are around, how often did they change the law? This metric answers that question.

PI (Potential Impact): The problem with just looking at the marginal vote is that it treats someone in a four-justice minority the same as someone who dissents alone. This stat is calculated as the reciprocal of the number of justices, including this justice, who would need to change their mind for this justice’s vote to have marginal impact. It captures how a justice might’ve been impactful, if the other justices were different, weighted by how different the other justices would’ve needed to be.

PI/VC: Potential impact per vote cast.

IW (Issues Written): Note that this measure is of issues written, not reasons written. A set of reasons may address multiple issues. This metric is a way of seeing whose drafting pen had the most impact on Canadian law (as opposed to whose signature pen).

VF (Votes For): The total number of votes a writer received. A measure of how many times a justice convinced another justice of the Supreme Court that they were right when they decided to write.

VF/IW: A measure of, on average, how many other justices agreed with the writing justice. Note that this metric treats writing a unanimous opinion for a five-justice panel as the same as writing a five-justice majority opinion for a nine-justice panel.

%(VF): The average percent of votes available that a justice received when writing. Writing a unanimous opinion on a five-justice panel counts as much as writing a unanimous opinion on a nine-justice panel.


(db last updated: 2022-02-11)