Not Taking Any Chances


By David Rangaviz

State v. Vuley, 2013 VT 9

“The man who wins the lottery once is envied; the one who wins it twice is investigated.”

In criminal trials, the presentation of evidence to the jury is governed by certain rules.  These evidentiary rules often keep highly relevant information away from the jury, typically because the evidence is either somehow unreliable, unfair to the defendant, or confusing to the jury.

One such rule bars what is referred to either as propensity or character evidence.  Under this rule, the prosecution cannot admit evidence of a defendant’s other bad acts—as in, not the act the defendant is on trial for allegedly committing—to show that he has a certain character trait or propensity (for committing crime).  The theory behind the rule is that the prosecution should have to show that the defendant committed this crime, not have the bar effectively lowered by the admission of evidence of the defendant’s prior bad acts.  Also, these prior bad acts might poison the jury against the defendant, and send the trial down any number of rabbit holes if the defendant disputes that he committed each individual prior act (turning each trial into an endless criminal version of This Is Your Life).


This ban on character evidence is also frustrating because it contradicts another fundamental purpose of the evidentiary rules: putting the best (i.e., most relevant) evidence before the jury.  It’s obviously highly relevant to a bank robbery trial that a defendant has committed five prior bank robberies; bank robbers rob banks.  That’s why we rely on this type of reasoning all the time in our everyday lives.  What’s past is prologue.

But the strength of this evidence is exactly why it’s banned.  The inference of propensity overwhelms the other evidence and the jury may convict on this basis alone.  In other words, the evidence is so powerfully relevant that it’s unfair.  (Besides, bank robbers don’t always rob banks.  Everybody can change!)

This tension means the stakes of such cases are high.  Admit the evidence, and the defendant will likely be convicted.  Keep it out, and he may walk free.  This makes the character evidence prohibition one of the most heavily litigated provisions of the rules of evidence.

Today’s case presents an interesting wrinkle on this commonplace evidentiary issue—should a jury be allowed to infer a defendant’s criminal intent based on the improbability of multiple coincidences?

Let me explain.

Over the course of eight weeks, from late 2008 to early 2009, four fires occurred at Defendant’s house in Colchester.  That’s one fire every two weeks!  To give you some context, in all of 2008 and 2009 the Malletts Bay Fire Department responded to ten fires.  Four were at Defendant’s house.

Really really (really) bad luck?  Or was there something a little more nefarious going on here?

The State’s Attorney thought the latter and charged Defendant with four counts of first-degree arson.  During the trial, the judge instructed the jury that, based upon their common sense, they could conclude “that it is unlikely that this number of similar accidents would occur on the same property in a short period of time.”

Defendant was convicted of setting the final two fires.  He appealed to SCOV.

In an extraordinarily detailed opinion, the SCOV unanimously holds that the jury instruction, and the evidence of all four fires, permits the very same propensity reasoning barred by the rules of evidence.  As such, the instruction was erroneous.

The doctrine in question—allowing the jury to infer that the four fires were not accidental—is referred to as the “doctrine of chances.”  According to SCOV, the use of this evidence allows the jury to make three distinct inferences.

First, the “probabilistic” doctrine of chances allows the jury to infer that, where independently improbable events occur together, it is likely that some common cause explains them all.  If lightning strikes the same place ten times, there’s probably a reason other than mere chance.

Second, when this doctrine is applied to human conduct, it allows the inference that independently improbable incidents that might have been accidental were not, in fact, accidents.  If I shoot ten of my wives and claim that all ten shootings were accidents, there’s likely a reason other than accident that explains the shootings—I’m a murderer.  (Hopefully, I would have been caught earlier on in this Bluebeard scenario.)

Third, the “psychological” doctrine of chances permits the inference that when something unusual has already happened, subsequent occurrences are less likely to be accidents.  If I actually shoot my first wife by accident, I’d probably learn from that mistake and be much less likely than the average man to shoot my second wife accidentally.  (To take a real world example: If it really was a mistake, Oscar Pistorius probably won’t do it again.)

Now a quick quiz.  Which, if any, of these three inferences sounds like impermissible propensity reasoning?

According to the SCOV, the answer is inference #2.  By the court’s description, this inference is simply covert propensity reasoning—it allows a jury to infer that “because a man has intentionally killed a wife, he is therefore more likely to have intentionally killed this wife.”  To assume that all of the fires derived from the same source, we would have to assign Defendant the character trait of being an arsonist, exactly the sort of propensity inference barred by the rules.

As SCOV recognizes, many commentators and courts disagree, arguing that “chances” evidence is not really propensity evidence.  Instead, so this argument goes, the evidence is really about objective reality, not subjective character.  There is simply a disparity between the statistical odds that an innocent person would have four fires at their home and the odds that a guilty person would have so many fires.  The jury should be able to hear this evidence and rely upon this objective truth in assessing guilt.

But SCOV isn’t fooled.  This is pure propensity reasoning in statistician’s clothing, as this observation could only be true if and only if the guilty person has the propensity to set repeated fires.  SCOV says it better than I ever could: “Propensity reasoning does not become permissible by virtue of being backed by objective probabilities.  The objective/subjection distinction is a red herring here.”

As a result, such “chance” evidence is inadmissible because it allows the jury to engage in improper propensity-based reasoning.

The difficulty here lies in the application of the exceptions to the general prohibition on character evidence.  The rules expressly permit character evidence where it is used to show intent or the absence of an accident.  That exception would seem to apply here, and allow the admission of the previous fires to show that Defendant intentionally set the fires and they were not accidents.

If you’ve followed along closely, you might see that this is the SCOV’s “psychological” doctrine of chances (inference #3 – Oscar Pistorius learning from his mistake).  The fact that an unusual incident occurred means that a future recurrence will be less likely to be an accident (and thus more likely to be intentional).  According to the SCOV, this inference is permitted because it is “more about the relevance of individual history to a person’s mental state.”

But one question the SCOV doesn’t answer is whether prosecutors will be able to take advantage of this third kind of inference in future cases by use of a limiting instruction.

Limiting instructions allow juries to hear inadmissible evidence if, upon hearing that evidence, they are also admonished by the judge not to consider the evidence for the unauthorized purpose and to only use it for permissible purposes.  In our jury system, courts assume that jurors follow these instructions.

Applying this common practice, it is certainly possibly that in some future case a Vermont trial court might admit such “chance” evidence while giving a limiting instruction disallowing inference #2 but allowing inference #3.  The practical difference, then, might be negligible—the jury can hear the “chance” evidence while just being told that they shouldn’t consider it in a certain way.  Even a highly-sophisticated jury would have a hard time telling the difference.

For now though, the SCOV has it just right.  Propensity evidence is so powerful that courts must carefully police its admission.  The SCOV has struck the proper balance.

It was a hollow victory for this Defendant, however, as SCOV affirms his arson convictions because his attorney failed to object to the jury instruction properly, thus effectively waiving the error during the trial.  (Justice Robinson dissented on this point alone.)

So the next time you try to predict someone’s conduct by assessing their character, just remember that your inference, while commonplace throughout society, wouldn’t be allowed in a courtroom.  

Also, if you happen to have really bad luck, just try never to be unlucky in the same way twice.

Because, seriously … what are the chances?

Comments