Prop 8 and the Issue/Outcome Voting Problem

By Mike Dorf


I haven't yet finished watching the video of the Ninth Circuit Prop 8 oral argument.  (In fact I've just started.)  I may blog about one or more tidbits in it later in the week.  For now, I want to use the case to explore the problem of issue voting versus outcome voting.  This is a problem that is almost always potentially present for a multi-judge court addressing multiple issues, so there's nothing really special about Perry v. Schwarzenegger here, but the case does present a nice vehicle for introducing readers to the problem.

On appeal, Perry  presents multiple issues and sub-issues but for our purposes we can simplify it as follows:

1) Is there an Article III case or controversy on appeal such that the court's exercise of jurisdiction is consistent with the Constitution?

2) On the merits, should the district court's decision invalidating Prop 8 be reversed?

In order for the court to reverse the district court ruling (i.e., in order for Prop 8 to be validly enforceable), the answer to both questions must be "yes."  Among the complexities I do not address here is whether a decision that there is no appellate standing would entail that there was no case or controversy in the trial court, in which case the Ninth Circuit might have to say that the district court never should have entertained the case, and dismiss for that reason.  But I'll put that aside and assume that issue 1 is simply a question of appellate jurisdiction.

There are three judges: Reinhardt, Hawkins, and Smith.  Let's suppose they vote as follows:

Reinhardt: 1) Yes; 2) No

Hawkins:  1) No; 2) Yes

Smith:  1) Yes; 2) Yes

What is the bottom line here? If we engage in what the literature calls "outcome" voting, whereby we add up the bottom-line votes of each judge, we have two votes to affirm (Reinhardt and Hawkins), and so the district court judgment is affirmed.  But if we aggregate the judges' votes issue-by-issue, we get the opposite result.  The judges say 2-1 that there is appellate jurisdiction, and they also say 2-1 that the district court got the merits wrong.

Issue voting feels like the right way to go here.  One way to think about what makes outcome voting so peculiar is to ask what would happen in subsequent cases in which the appellate standing issue and the merits issue are not presented together.  Would Perry stand for the bottom line reached on each issue?  Or would the votes on those issues simply be dicta because--if we are using issue voting rather than outcome voting--they did not play a necessary role in supporting the result?

Note that if courts were really committed to outcome voting, then in this example Judge Hawkins wouldn't even register a view on question 2.  Having found no jurisdiction, it would be unnecessary for him to address the merits.

What is the right answer to the aggregation question?  Perhaps surprisingly, the issue is not exactly resolved.  Where, as in this particular example, one has a jurisdictional issue and then a merits issue, courts typically engage in issue voting.  Thus a judge who thinks there is no jurisdiction but is outvoted on that question, then gets to register a vote on the merits.  But where a party needs to win on two separate merits issues in order to win the case--as is often true of parties challenging lower court rulings that may be defended on alternate grounds--the usual convention is in favor of outcome voting rather than issue voting, even though that leads to paradoxes.

Here's another example using two merits issues.  Suppose that at D's murder trial he moves to suppress a bloody knife on the ground that it was the product of an unlawful search and he also moves to suppress a confession on the ground that he did not properly receive Miranda warnings.  The trial judge denies both motions, the evidence is admitted, and D is convicted.  He appeals to a panel consisting of Judges Tom, Dick and Harry.  They vote as follows:

Tom:  Knife admissible; Confession inadmissible (and not harmless error)
Dick: Knife inadmissible (and not harmless error); Confession admissible
Harry: Knife admissible; Confession admissible

Here two judges have voted to reverse and so outcome voting means the conviction is reversed.  But the ruling is 2-1 in favor of admitting the knife and also 2-1 in favor of admitting the confession.  In this instance, at least to me, outcome voting seems like the right result, although that raises the same problems that it does in Perry.  For example, what happens on remand for a new trial?  When D again moves to suppress the knife and the confession, is the district judge free to admit one or the other but not both (because admitting just the knife or just the confession will not produce a reversal, but admitting both will--assuming a conviction results)?  How should he choose?

And what exactly is the difference between this case and the example above based on Perry?  I don't think the fact that the knife/confession case is a criminal matter explains the different intuition.  I have the same intuition if a civil case involved challenges to the admissibility of two pieces of evidence in this way.  The fact that one of the issues in Perry is jurisdictional does some work, but there too, one shouldn't get hung up on labels.  Often the question of whether some issue even is jurisdictional or not can be difficult.

I find these puzzles fascinating and wonder why they so rarely materialize.  I strongly suspect it's because appellate judges are deeply pragmatic.  They typically work together to generate coherent outcomes as a group, even when they disagree about how each judge would get to the bottom line if deciding cases as an individual.

Finally, I want to acknowledge that I haven't said anything especially original here.  The leading paper is by Lewis Kornhauser and Larry Sager, Unpacking the Court, in the Yale Law Journal in 1986.  A somewhat more recent Philosophy & Public Affairs paper by Christian List and Philip Pettit succinctly discusses some of these issues.