Multi-Relevant Evidence in Criminal Cases

by Sherry F. Colb

In my column for this week, I discuss the case of Buck v. Davis, in which the Supreme Court, after cutting through a procedural morass, ruled that a capital defendant's attorney had been constitutionally ineffective under the Sixth Amendment when he introduced an expert witness who said that the defendant, though unlikely to be dangerous in the future, did have one characteristic (his race) that statistically correlated positively with future violence, a correlation that could have led the jury to sentence the defendant to death (as the jury was permitted to do only if it found that he was likely to be violent in the future).

In my column, I discuss the importance of race to this case, observing that ordinarily, if an expert witness has something helpful to say (i.e., that the client is not dangerous) and something harmful to say (i.e., that the client has a characteristic that correlates positively with dangerousness), it is acceptable practice for the attorney to decide it is worth introducing the expert witness's testimony, notwithstanding the harmful portions of that testimony.  In other words, evidence is generally not wrapped up nicely in a bow of purely helpful testimony that a defense attorney can choose to utilize without any downsides.  This is where strategy comes in, and a lawyer makes a judgment about whether the helpful aspects of the testimony do or do not outweigh the harm that it might do.  It was because the expert in Buck was feeding a virulent racial stereotype about violent black men that it made sense for the Court to find that the attorney who presented the expert was ineffective, having made an inexcusable judgment call, and that the error likely had an impact on the outcome of the case.

In this post, I want to observe that the phenomenon of the "mixed bag" in evidence is hardly unique to the context of an attorney deciding whether to introduce an expert witness.  As I teach my students in Evidence class, even though the various rules of evidence tend to offer bases for excluding rather than for admitting evidence, most of the rules prohibit inferences rather than evidence.  This means that a typical rule will not say that all evidence (of liability insurance, for example) is inadmissible but will say instead that such evidence may not be offered to support a prohibited inference (of negligence, for instance).  To the extent that the evidence is relevant only to prove the point on which it is inadmissible, the evidence stays out.

However, in the many circumstances in which the evidence is relevant both on an impermissible point and on a permissible one, there is a good chance that a judge will choose to overrule an objection to the evidence and admit it subject to a limiting instruction telling the jury that it may not rely on the evidence to prove the impermissible point but may use it only as proof of the permissible conclusion.  This process resembles the efforts of a defense attorney to capitalize on the helpful testimony of an expert and encourage the jury to focus on that rather than on the harmful aspects of the testimony.

Unfortunately, it is rare to have evidence that is pure and relevant only on a point on which it is admissible.  More commonly, evidence will have multiple relevances.  For example, a criminal defendant's prior conviction for grand larceny is relevant both to prove that he is less trustworthy as a witness than he would otherwise be (credibility) and that he is more likely to have committed the robbery of which he is charged than he would be in the absence of this evidence (guilt).  The prior conviction is admissible, however, only on the issue of credibility, not as proof of the defendant's guilt.  The compromise that the rules of evidence often provide is that the conviction may come in subject to a limiting instruction telling the jury not to use the conviction as proof of guilt but only as proof of (the lack of) credibility.

It turns out, based on empirical studies, that jurors have a very difficult time limiting their consideration of evidence to the purposes for which they have been instructed to consider it. Especially if evidence is highly relevant to something for which it is inadmissible, jurors will tend to pay attention to it (and perhaps pay even greater attention when they have heard a limiting instruction telling them not to draw the impermissible inference).  Like the jurors who heard from the expert both that the defendant in Buck was not dangerous and that his race statistically correlated with future danger, jurors hearing evidence with multiple relevances will likely attend to everything, not just to the part that is permissible/helpful.

For the law of evidence, this is where the important job of Federal Rule 403 comes into play.  Under Rule 403, if the probative value of evidence is substantially outweighed by its tendency to cause unfair prejudice (or other problems, such as juror confusion), the evidence may be excluded altogether, despite its relevance to a point on which it is admissible. Through Rule 403 (and its analogues in various rules that modify which way the balance cuts), it becomes possible to take into account the reality that juries--at least sometimes--cannot follow the limiting instructions that they receive and that the best course may be to keep out the evidence altogether.  For the evidence of racially-correlated dangerousness, this too would almost certainly have been the right call.