I. GETTING CHARACTER EVIDENCE THROUGH THE COURTROOM DOOR

It is tempting to give in to Laudan's critique of current character evidence rules, especially his attack on the asymmetry in those rules which allows the defense to introduce evidence of the defendant's good character but disallows the prosecution doing so with regard to the defendant's bad character, except in rebuttal. If evidence of good character is “relevant to, without being dispositive of,” our evaluation of people's conduct, why should not evidence of bad character be admissible?Footnote ¹⁰ But before we accept this, it is important to get clearer on just what character is and what the various kinds of evidence that might be introduced at a criminal trial reveal about it. It is undeniable that we use character evidence in our everyday lives to predict people's behavior or interpret it when their motives or intentions are not otherwise transparent. Yet the contexts in which we do so and the stakes involved are often very different from those of a criminal trial. Moreover, the kinds of evidence that we have of the character of individuals are often superior to those juries are apt to have.

To begin with, we need some workable account of what “character” is, for this will help us see how the debate about character evidence is, at times, misleadingly cast. Joel Kupperman offers an account of character according to which “X's character is X's normal pattern of thought and action, especially with respect to concerns and commitments in matters affecting the happiness of others or of X, and most especially in relation to moral choices.”Footnote ¹¹ Kupperman notes that by “normal pattern,” he means what is normal for X in various circumstances, including unusual ones that might be regarded as moral test cases for X. There is an implication here of considerable stability. In speaking about a person's character, we pick out things about a person that are not momentary concerns but are instead ones that “bridge the present and other parts of a person's life” and also “bridge thought and action.”Footnote ¹² This is not to say that character cannot change over time. It can, and this fact will have some relevance to the use of past criminal records in trials.

Consider also Christine McKinnon's slightly different account: “To talk about a person's character is to talk about a complex of virtues and vices and more ethically neutral skills, abilities, and dispositions she possesses as well as to talk about the ways in which she values them and identifies with them and the roles she has constructed for them in her life.”Footnote ¹³ McKinnon's account, like Kupperman's, emphasizes the way in which character unifies a life, because it involves a somewhat self-reflective stance on one's dispositions and their desirability. Thus, when we say that someone has good character, we mean that she not only stably possesses certain kinds of moral sensibilities and skills that enable her to act well but also that she is committed to developing and extending them and to pursuing aims and purposes that are consistent with the dignity of others and proper concern for their welfare. By contrast, when we say that someone has bad character, we mean that she lacks crucial moral sensibilities and skills, has little interest in extending them or eschews the value of doing so, and tends to pursue projects or aims that show little regard for others' dignity or welfare.

One thing that should be readily apparent from these accounts is just how much we would need to know about a person in order to have something approaching justified beliefs about his character. Though we can read a good deal about a person's character from his actions, really knowing his character requires delving beneath his surface conduct to get at his motives, concerns, ways of thinking, and attitudes toward himself and others. Typically, we only come to “know” a person's character after spending considerable time observing him in a variety of different contexts, talking with him, or talking to others who have observed and interacted with him over an extended period of time. Of course, we may quickly form provisional judgments of character because we have to in order to interpret people's behavior or predict what they will do. But such provisional judgments are properly subject to revision and refinement in light of further observations of or interactions with the individuals in question. Obviously, such revision and refinement can be an ongoing process.

In our everyday affairs, we employ character judgments of both provisional and more fully evidenced kinds. But this may not tell us much about whether doing so in the context of criminal trials is feasible or even advisable. For one thing, the evidence of a defendant's character that juries are apt to see during a criminal trial is sometimes of dubious quality. This will certainly be true for so-called “reputational testimony,” where witnesses come before the court to state their views about a defendant's perceived standing in the community. As Uviller puts it, the witness most likely “merely projects his personal opinion into the hypothetical minds and conversations of an ill-defined group of anonymous neighbors and associates.”Footnote ¹⁴ How reliable such testimony will be is anyone's guess, but there seems little reason to believe that it will usually be based on a witness's careful observation of and interaction with the defendant in a variety of contexts over an extended period of time. If such testimony is supposed to help jurors decide whether the defendant is the kind of person who would commit the crime or crimes with which he is charged, it seems very weakly probative at best and is properly subject to contestation.

What about character testimony offered by witnesses who are called upon to give their personal opinion of a defendant's character? Such testimony might be more probative, depending on the extent and kind of interactions the witness has had with the defendant as well as upon the witness's skill at observing and interpreting people's behavior. In some cases, such testimony might give us quite a bit of useful evidence about a person's character. Yet what it will more often give us is evidence about certain limited patterns of past conduct by the defendant. Indeed, these are the sorts of things witnesses can probably be relied upon to attest to—specific instances in which they have observed the defendant acting in certain ways in the past. If, for instance, a defendant is charged with recklessly endangering the lives of others in some specific type of context (e.g., while driving an automobile), then having witnesses who testify to having observed the defendant behaving in similar ways in similar contexts in the past seems probative. How much it should be allowed to count toward establishing the guilt of defendants BARD is a matter I return to below. The important point here is that though such past misconduct may suggest something about the defendant's character, it would be hazardous, to say the least, to draw any broad or firm conclusions about the defendant's character from testimony about it.

It might be objected that particular bits of trace evidence will often be no more than weakly probative, pointing toward the defendant's guilt in only a mildly probabilistic fashion. Yet they are offered up to the jury as part of the prosecution's larger case against the defendant. The jury is expected to sort through the various kinds of trace evidence proffered and arrive at a conclusion about its overall strength. Why not allow marginally probative character evidence to be introduced and evaluated by the jury as well, especially if it is subject to vigorous challenge by defense attorneys? This objection is well taken when reputational and personal opinion testimony are at issue but should be resisted when it is a defendant's criminal record that the prosecution seeks to introduce. Why is this? In the first place, a person's past criminal record may not tell us much about his overall character or even the more specific propensities of thought and action that he currently possesses. As such, his criminal record will often be only very weakly probative. In the second place, the likelihood that introduction of a defendant's past criminal record at his trial will bias the jurors against him, independently of the relative strength of the trace evidence in the case, distinguishes it from reputational or personal opinion testimony. Possession of a criminal record, especially one that is lengthy or violent, is highly stigmatizing. Indeed, there is evidence confirming the tendency of jurors more readily to convict defendants about whose criminal pasts they learn.Footnote ¹⁵ It therefore seems reasonable to exclude evidence of a defendant's past record unless it can be shown to provide specific evidence about his current propensities.

We should be quite hesitant in making inferences from a defendant's criminal record to claims about her character. A person's past criminal record does not “speak for itself,” especially insofar as it is consulted for evidence about her character. Set to one side, for the moment, defendants who have been wrongly convicted in the past or who have pled guilty to crimes that they did not commit because offered attractive plea bargains that they accepted because they did not want to risk conviction at trial, with its typically harsher sanctions. Suppose that a defendant is guilty of all of the crimes of which she has been convicted. Her record may still tell us little about the circumstances of and motivations behind her offending. Did she commit these crimes because she is self-centered and violent or because she was addicted to drugs or desperately attempting to please a violent man on whom she was economically or emotionally dependent? If a history of criminal offending is to provide us with character evidence, then we must find out more about the origins of her conduct to discover what (if anything) her record reveals about her normal patterns of thought and action. This suggests that the courts would have to take on the costly and time-consuming tasks of finding out not just what defendants have done in the past but why they have done it—for what purposes, in what contexts, and in what ways. For it is only these details that reveal a defendant's relatively stable propensities for thought and action, her responsiveness (or lack thereof) to moral considerations, and thus her degree of concern for the dignity and welfare of herself and others.

Furthermore, a defendant's past criminal record might be “stale” as an indicator of her character. A series of convictions five or ten years ago may not tell us much about a defendant's current propensities, at least on the plausible assumption that people can gradually change and thus not be the kinds of people they were at earlier junctures in their lives. Perhaps a defendant's attorney can be trusted to raise such doubts and complexities, but it is apparent that they should be raised.

Admittedly, there will be cases in which a defendant's criminal record does reveal things about his current character. A long, diverse, and violent enough criminal history (one the validity of which is beyond reasonable dispute) might give us quite a bit of useful evidence about a defendant—in particular, his relatively stable willingness to run roughshod over others' interests in the pursuit of his own. Such evidence might be bolstered by the testimony of family members, friends, or victims of prior crimes who have directly witnessed the defendant's depravity in a wide variety of contexts. Moreover, if a number of the convictions are quite recent, there will be little reason to believe that the defendant has changed his stripes. Even in such cases, however, we should be cautious about drawing sweeping conclusions about a defendant's character. We all know that humans are complex creatures. Depraved killers are sometimes kind to their friends, mothers, or pets. Hence a past criminal record is unlikely to provide us with a complete or nuanced portrait of a defendant's character. Instead, it might give us useful information about only slices or segments of his character or about specific propensities that he possesses.

Perhaps that is all we need, since it might plausibly be suggested that what we are attempting to uncover with such evidence is whether defendants are the kinds of persons likely to commit the crimes with which they are charged, especially if given the opportunity to do so. Defendants' past criminal records might tell us this much, if not all that much about their broader characters. True enough, though it is important to emphasize the difference between more and less specific kinds of such evidence. It is one thing if a past criminal record provides credible evidence that a defendant is the kind of person likely to commit the specific type of crime with which he is currently charged. It is quite another if a past criminal record provides credible evidence that a defendant has more general criminal tendencies that are less directly related to the specific crimes with which he is currently on trial. The first kind of character evidence might be significantly probative and enough so to convince us to set aside our concerns that its introduction will bias jurors against defendants so much that they will discount or ignore the relevant trace evidence. The second kind of character evidence is more problematic, and not just because people can act contrary to their tendencies or because such tendencies must not be permitted to overshadow the trace evidence of guilt. A tendency to commit one type of crime may reveal little about the tendency to commit other types. True, the evidence we have from recidivism studies suggests that many offenders are generalists when it comes to committing crimes.Footnote ¹⁶ Most do not specialize. But it would be hazardous, to say the least, to infer present guilt with regard to one alleged type of offense from a past record of other kinds of offenses. Hence I would disallow the introduction of a past record that is unrelated to the current charges. Such evidence is not probative enough to warrant our taking the chance that its introduction will overshadow the trace evidence of a defendant's guilt in jurors' minds.

Notice that it is the specificity of certain kinds of past-conduct evidence that explains its undisputed probative value in criminal cases. As already noted, evidence rules permit the prosecution to introduce evidence of a defendant's past misconduct: a modus operandi used by the defendant in the past for which there is trace evidence in the current case; evidence that the defendant had a motive to commit the crime with which he is currently charged; or evidence (such as a past conviction for the same offense) that the defendant knows how to commit a specific kind of offense with which he is currently charged (e.g., pickpocketing), especially if the defendant actively disputes this contention. Such past-conduct evidence is “specific” in the sense that it establishes an element of the current charges (e.g., knowledge of how to commit the crime in question), or corresponds with an element of the current charges (e.g., a modus operandi) or an overall crime pattern exhibited in the current charges, or rebuts all or some part of a defense offered by a defendant. A.E. Acorn argues that many of the more specific kinds of past-conduct evidence are probative precisely because they do not ask juries to reason from broad and somewhat dubious generalizations about people (e.g., people who have past criminal records are bad people and thus likely to be guilty of the current charges against them).Footnote ¹⁷ Instead they invite juries to reason from more specific types of generalizations (e.g., people who have been repeatedly convicted of crimes involving a specific modus operandi, which is also present in the current case, are more likely to be guilty of the current charges). Hence such past-conduct evidence is not employed by the prosecution to implicate the defendant based on his bad character. Indeed, such evidence does not tell us much at all about the defendant's character so much as provide or sustain trace evidence.Footnote ¹⁸

Assuming that certain kinds of character evidence are deemed probative enough to rebut the presumption against allowing defendants' criminal records to be introduced into a trial, there should be some procedure for determining when doing so is to be permitted. It seems that the burden of establishing the freshness, reliability, and specificity of the evidence should be on the prosecution, and we might reasonably require the prosecution to provide clear and convincing evidence of this freshness, reliability, and specificity to the presiding judge. Defense counsel should be allowed to challenge the introduction of such evidence along any of the relevant dimensions, arguing that it is dated, unreliable for indicating any sort of propensity of the defendant, or not specific enough given the current charges and evidence for them. Also, the decision by a judge to permit the introduction of a defendant's past criminal record should be a basis for subsequent appeal if the defendant is convicted. Importantly, any hearing on the introduction of such character evidence should be conducted away from the jury, so that jurors do not get wind of the defendant's past criminal record until or unless its introduction into evidence is allowed by the judge.Footnote ¹⁹ If it is allowed, defense counsel should be permitted to contest it vigorously, much as they would any other evidence, and caution jurors about the inferences to be drawn from it. Even this may not be enough for, as we see below, jurors should evaluate the strength of the trace evidence independently of probative character evidence against the defendant.

This procedure for introducing character evidence, as well as the relevant criteria of freshness, specificity, and reliability, arguably should also be employed when efforts are made to impeach witness testimony, especially when the witness in question is the defendant herself. It should not be permissible for the prosecution to seek to impeach the defendant's testimony (and at the same time prejudice the jurors against her) by raising her past criminal record should she choose to testify. Of course, if her past criminal history includes one or more convictions for perjury, that might be deemed sufficiently probative by the judge to overcome the presumption against introduction of her criminal record in response to her testimony. Similarly, should she deny any motive to commit the crime in question or knowledge of how to commit it, then evidence from her past criminal history might be reliable, fresh, and specific enough to warrant a judge's allowing it to be introduced (subject, of course, to contestation by defense counsel). Indeed, even in cases where the witness to be impeached is not the defendant herself, I am not convinced that quite general and vague attacks on the witness's character, of the sort that derive from citing the witness's criminal record, should be permitted.Footnote ²⁰ Nonetheless, I concede that such cases are more complicated because the witness is not a criminal defendant whose fate is in the jury's hands.

There are, no doubt, harder cases where the past-conduct evidence that the prosecution seeks to introduce will not be so general as to utterly fail to show the defendant to be the kind of person who might commit the crime in question nor so specific as to establish an element of the current offense or demonstrate a modus operandi or distinctive crime pattern. The number of previous convictions a defendant has can vary, as can their temporal distance from the present or extent to which they have elements that bear on the current charges. Where to draw the line on the admissibility of less reliable or specific past-conduct evidence is a difficult question. Under ideal conditions, we might err on the side of allowing more rather than less of it to come before judges or juries, with the reasonable hope that they can properly sift through and weigh it. Under less-than-ideal conditions, we might be more reluctant to trust those who must decide the fate of defendants with less reliable or specific past-conduct evidence. I return to these matters later in the discussion.

II. THE RELATIONSHIP BETWEEN TRACE AND CHARACTER EVIDENCE

In this section, I argue that even if the courts do permit the introduction of reliable, fresh, and specific character evidence regarding a defendant, jurors should not be encouraged to rely on it very much in making decisions about guilt or innocence. In other words, the admissibility of such evidence should not be taken as anywhere near sufficient to establishing the truth of the current charges against the defendant. Again, what such evidence shows is that the defendant is the kind of person who is disposed to commit the crime with which he is charged. It obviously does not show that he has done so on the occasion in question. It is the trace evidence that can show that, though I concede that character evidence might have a legitimate role to play in supplementing the trace evidence in some instances.

It is vitally important to us to come to understand what makes others tick, not only to predict their future behavior but also to figure out how to approach and deal with them. Especially when our interpreting or predicting others' behavior has relatively low stakes, we seem justified in relying on character evidence, even provisional character evidence. Yet notice how things change when the stakes are higher. Suppose that one's supervisor has been accused of the wrongful firing of a colleague. Suppose also that one has credible evidence concerning the supervisor's good character, evidence that justifiably leads one to believe that it is highly unlikely that he would engage in misconduct of the sort in question. One has observed his conduct over a long period of time in diverse situations. In addition, numerous conversations with him suggest that he is stably disposed to act conscientiously and from the right sorts of reasons. Other colleagues whose judgment one believes is reliable have reached similar conclusions about his good character, thus corroborating the character evidence one has. Nonetheless, given the stakes in the present case, the character evidence, credible as it may be, is not at all decisive. If the allegations against the supervisor are serious enough and the impact on affected parties significant, then it seems reasonable, if not mandatory, to insist upon a more formal inquiry into his conduct.

Arguably, in the course of such a formal inquiry, character evidence should be temporarily bracketed, and the relevant trace evidence of his conduct with regard to the matter in question should be given center stage. The suggestion that strong evidence of his good character should be employed to examine and interpret the trace evidence in a light most favorable to him should, it seems, be resisted. If we are to evaluate the trace evidence impartially, we should be prepared to view it in ways that are both maximally charitable and uncharitable to him. Only then will we be in a position to draw reasonable conclusions about which interpretation of his conduct the trace evidence best supports. Indeed, one of the reasons for bringing in third parties to adjudicate such disputes is precisely that they are less likely to have their evaluations of the evidence colored by what they know about the accused. If the trace evidence of misconduct is of the right kind, even the best character evidence may be rebutted. The explanation for this is the familiar one that people with good character can sometimes behave badly. At most, credible evidence of the supervisor's good character might come into play if the trace evidence, evaluated from all sides, so to speak, is suggestive of misconduct but not conclusive. At that point we might reasonably allow in the character evidence to reach the conclusion that he likely did not engage in the misconduct of which he is accused. Of course, if we lack credible character evidence, our conclusion might have to be the more cautious one that the case against him is unproven.

The supervisor case is one that falls into the first category of cases that I distinguish above, since the question it raises is whether the accused individual engaged in misconduct. There are no other suspects in the case, in the sense of other individuals who might have wrongfully fired the colleague. Similarly, juries in criminal cases must sometimes decide whether an accused person acted criminally, not whether he is the person who acted criminally. We might refer to the former as “identity” cases, and the latter as “nonidentity” cases.Footnote ²¹ To what extent or in what ways should credible character evidence, assuming it is available, be deemed useful in deciding the nonidentity cases?

The answer, it seems, depends on both the quality of the trace evidence in the case and the relative specificity of the credible character evidence. Suppose that a defendant is on trial for rape. His defense is that the complainant consented to the sexual intercourse, a claim that she vehemently and convincingly denies when put on the witness stand. Yet she was seen kissing and hugging the defendant in public, and she admits to going to his apartment willingly on the evening in question. For his part, the accused is a quite credible witness on his own behalf when put on the stand. Suppose that there is no evidence of a motive on the complainant's part that would cast doubt on her version of events. Even if we think that woman should be believed in such cases, that it is highly unlikely that they would go to the trouble and potential embarrassment of publicly accusing someone of raping them if it were not true, it is also true that the evidence, as described, may not be BARD conclusive of the defendant's guilt. Perhaps a preponderance of the evidence (or PoE, to use Laudan's acronym) supports a conclusion that the defendant is guilty. In other words, there are more reasons for judging him guilty than for judging him not guilty, or the reasons for judging him guilty are stronger than those for judging him not guilty. Or if the presumption that women generally do not formally accuse men of rape and testify against them in a court of law unless they have been raped is granted, we might conclude that there is clear and convincing evidence (or CACE, to use Laudan's acronym) of the defendant's guilt in the case. The CACE standard should be understood as saying that there are many more or stronger reasons for judging him guilty than not. Still, the standard of proof in criminal cases is higher—clear and convincing evidence that the defendant is guilty and rebuttal by the prosecution of any evidenced alternative account of the incident in question according to which the defendant should be judged not guilty.Footnote ²² There is, in the case as described, an alternative explanation of the events according to which the defendant is not guilty (namely, that the complainant consented to sexual intercourse), though how well evidenced it is admittedly seems unclear.

Suppose, however, that the prosecution had and could introduce fresh, reliable, and specific character evidence that the defendant is the kind of person who commits violent sexual assaults against women, only to subsequently claim that they consented to the sexual intercourse.Footnote ²³ Again, we may not often have such evidence based on a defendant's past criminal record, but suppose, for the sake of argument, that we did in a given case. It is hard to resist the contention that the jury should be given such evidence, though they must be cautioned strongly about its use. True, such evidence might lead them to leap too quickly to the conclusion that the defendant is guilty. We would still want them to weigh it carefully against or beside the trace evidence. But character evidence of the requisite kind does seem to have probative value, perhaps enough to “tip the scales” so as to justify a finding of guilt BARD against the accused, especially if the trace evidence against him is clear and convincing or close to it.

Why is this? The answer, I believe, is that in the kind of case in question, we are in effect asking the jury to decide whether the defendant acted “in character.” We know, of course, that people sometimes act out of character—good people do things that they normally would not and bad people refrain from misconduct that they would normally engage in. At the same time, we also know that people quite often, and perhaps usually, act in character. If we did not know this, there would be little point to the whole debate about character evidence. We must be wary about finding defendants with unsavory characters guilty, but in the case described, we are to suppose that we have, in addition, substantial trace evidence suggesting that the defendant did act in character. If the trace evidence gets the jury that far, then it does not seem inappropriate for the character evidence to seal the defendant's fate.

Granted, there is a risk in allowing the introduction of such evidence—namely, that innocent defendants will on occasion be found guilty. But there is also a risk if we do not allow its introduction—that guilty defendants will sometimes be found not guilty. If it is argued that we have (or ought to have) a much stronger preference to avoid the former than to avoid the latter, Laudan's retort seems convincing. We do strongly prefer to avoid punishing the innocent over not punishing the guilty. That is why we set the standard of proof at such a high level in criminal cases. Yet once the BARD standard is put in place, we should allow the introduction of all relevant evidence.Footnote ²⁴ Those opposed to allowing the introduction of fresh, reliable, and specific character evidence must convince us that it is irrelevant to determining the guilt or innocence of defendants no matter how carefully it has been vetted by the courts and evaluated, alongside relevant trace evidence, by the jury. And that seems a most unpromising view to defend.

Nonetheless, as the quality of the trace evidence diminishes, so should our confidence that probative character evidence can plug the gap, so to speak. I can imagine someone defending the continued use of fresh, specific, and reliable propensity evidence to warrant a finding of guilt BARD in a given case even if the trace evidence is significantly short of clear and convincing. But in such cases the possibility that the defendant may have acted contrary to his propensity should loom larger. After all, these will be cases in which there are some reasons for believing that the defendant did not act in character (though perhaps more or stronger reasons to believe that he did) or there is only suggestive trace evidence that he is guilty, but little more than that. My sense is that a finding of guilt in such cases depends rather too much on evidence concerning the defendant's unpleasant past and not enough on evidence concerning the conduct for which he is on trial.

Again, the extent to which character evidence is fresh, reliable, or specific will be matters of degree. Suppose, for instance, that while fresh and reliable, the character evidence the prosecution seeks to present shows only that a defendant is prone to acts of violence or predation, though not ones of the specific kind for which he is currently on trial. The prosecution might be able to show that the rape defendant is the kind of person who steals from others or engages in brawls but not that he is a sexual predator who in the past defended his actions with dubious claims of consent on the part of his victims. True, the character evidence suggests that the defendant is not wholly averse to illegally victimizing others and thus might be thought to be somewhat probative. Nevertheless, we should be more reluctant to allow its introduction. Why is this? Because if we do, jurors will be implicitly (if not explicitly) asked to make two worrisome inferences as opposed to one: First, because the defendant is the kind of person who commits other kinds of crimes, he is therefore the kind of person who would commit the crime in question, and second, because he is the kind of person who would commit the crime in question, he therefore is more likely to have committed it. The latter inference is already somewhat worrisome, and the former is more so. If the trace evidence is only up to the PoE standard, then the two propensity inferences forge an arguably quite tenuous link from the trace evidence to a conclusion of guilt BARD. It seems clear that if we admit such propensity evidence at all, we should insist that the trace evidence in the case meet the CACE standard. And as the character evidence becomes less fresh, less reliable, or less specifically related to the kind of crime with which the defendant is charged, even trace evidence meeting that standard arguably would not be sufficient to establish the defendant's guilt BARD.

It might be objected that it will be exceedingly difficult for judges to craft instructions enabling juries properly to understand and use character evidence that has been admitted due to its freshness, reliability, and specificity. Even if judges succeed in making vivid the distinction between trace and character evidence and explaining how the former must be clear and convincing before the latter can be properly put into play to reach a verdict, we might worry that the character evidence will displace or simply overwhelm the trace evidence in the jurors' minds. Of these two concerns, the latter seems the more difficult one to address. Judges must often formulate and give complex instructions to juries, and it is not clear that the challenges they face in helping jurors understand how to weigh and balance character evidence against trace evidence are different in kind. Whether jurors will follow the instructions provided and therefore keep character evidence in its proper place, so to speak, is something that might have to be monitored aggressively by judges. As I note below, we may need to encourage judges to scrutinize and reject verdicts that are not supported by clear and convincing trace evidence in cases in which character evidence is admitted.

This brings us to cases in which the issue before the court is whether it was the defendant rather than someone else who committed the crime with which the defendant is charged. These kinds of cases are probably more common than the nonidentity cases, and it may be that those generally opposed to the use of character evidence in criminal trials are thinking mainly about them. Why might fresh, reliable, and specific character evidence be seen as having less probative value in these types of cases? Because knowing (or reasonably believing) that the defendant is the kind of person who would commit the crime in question if given the opportunity tells us nothing about whether she is the person who committed the crime in this case. There may have been numerous individuals in the vicinity of the crime about whom it is true that they would have committed the crime in question if given the opportunity. It makes little difference to add that we have reason to believe that most people act in character most of the time. For that is also true of a wide range of people with criminal propensities. All that fresh, reliable, and specific character evidence tells us is that the prosecution has charged the right kind of person with the crime. In this regard, it is somewhat like trace evidence that shows that the crime was committed by someone with blue eyes. Assuming that the defendant has blue eyes, such evidence shows that the state has charged the right kind of defendant. Still, it is worth noting that character evidence showing the defendant to be the right kind of person to have committed the crime actually shows less, since the crime may have been committed by someone acting out of character or by someone with uneven or wavering character. And character evidence is strongly prejudicial, so we must be on guard against its introduction overwhelming the trace evidence in a given case.

A further complicating factor in identity cases is this: when the perpetrator of a crime is unknown, police tend to focus on individuals in the vicinity of the crime who have past criminal records. This practice of “rounding up the usual suspects” may affect the profile of cases that wind up going to trial.Footnote ²⁵ When the practice yields actually guilty individuals, they are likely to plead rather than go to trial. It is the materially innocent who may insist on trials in order to vindicate themselves. This means that in identity cases, a higher percentage of those on trial may in fact be wrongly accused, in part because of their past records. Jurors may be unaware of these skewing effects of police practices, though presumably they could be informed of them. In any case, the danger that the innocent have been rounded up is real and should be more troubling in identity cases.

Suppose, for instance, that defendant A has been charged with homicide, and there is clear and convincing trace evidence establishing her guilt. However, suppose also that the evidence is short of BARD conclusive because the defendant has provided an alibi that the prosecution has not been fully able to refute or because some of the physical evidence is consistent with another individual's having committed the crime. Would allowing the introduction of probative propensity evidence concerning the defendant seal her fate as it did in the nonidentity case? It seems not. Such evidence does nothing to rebut her alibi or counteract the recalcitrant physical evidence; it only shows her to be the right kind of suspect in the case. There may be numerous other people who are the right kinds of suspects, and propensity evidence against the defendant does nothing to eliminate them. Granted, most jurors, if presented with such propensity evidence in the kind of case described, might find the defendant guilty. But the question is whether they should do so—whether such evidence does anything at all to bridge the gap between CACE evidence and evidence BARD. It is hard to see how it does.

There might be some identity cases in which propensity evidence is more probative. Consider a case in which A and B are both suspected of having committed homicide and where the prosecution can establish not only that they are the only two who could have committed it but that only one of them did so. The problem for the courts is determining which of the two committed the crime. Suppose that the trace evidence strongly suggests that it is A rather than B who committed the homicide, so the state charges A and the case proceeds to trial. Suppose, however, that the trace evidence is somewhat short of being BARD convincing. B was, after all, at the scene of the homicide, and A testifies that it was B who did the killing. Yet the physical evidence points toward A, and B testifies that it was A who did the killing. Suppose also that there is fresh, reliable, and specific propensity evidence showing that A is the kind of person who is prone to violence against others, whereas nothing of the sort exists regarding B. In that case it does not seem unreasonable for a jury to employ the propensity evidence to reach the conclusion that A should be found guilty of the homicide. By hypothesis, other individuals with homicidal tendencies who might have been the killer have been ruled out, and both the trace and propensity evidence point rather decisively toward A.

But suppose that there is propensity evidence showing that B also has clear homicidal tendencies. Then what? It still seems that the jury might act reasonably in finding A guilty, assuming that it knows that either A or B must have done the killing. The physical evidence points toward A, as does the propensity evidence. The jury might be given pause by the propensity evidence against B, but it would likely reason (and plausibly so) that if one of the two did the killing, A is far more likely to have done so. It might even reasonably reach such a conclusion if the trace evidence against A is somewhat weaker than clear and convincing. Still, it is easy to imagine harder cases. For instance, suppose that the trace evidence does not point decisively or even very clearly toward either suspect, but the propensity evidence points exclusively toward one of them. We might predict that a jury in a trial for A would find A guilty if the propensity evidence against A were damning, on the reasoning that if one of the two committed homicide, A is the more likely culprit. The same prediction would hold if the probative propensity evidence were against B.

But would juries act responsibly in reaching these conclusions? That is a tough call. On a strict interpretation of the BARD standard, one would think not. There are two plausible suspects, the trace evidence is inconclusive, and all jurors have to go on is propensity evidence that tells them of a defendant's tendencies, but nothing more. Reasonable doubt seems to exist, in the sense that there is an alternative explanation of events exonerating the defendant (namely, the other suspect did the killing) that the state may not have successfully rebutted. Jurors might in fact convict the defendant against whom the character evidence points, but it is not apparent that they should do so. Of course, if the character evidence implicates both or neither of the defendants, then juries in their respective trials pretty clearly should find them not guilty in spite of the powerful motivations they will have to hang the crime on someone.

It is important to notice how things change if A and B are not the only two possible killers but the only two that the police and prosecution have identified as such and focused their energies upon. In that sort of case, clear and convincing evidence that A or B committed the homicide, even when combined with credible propensity evidence regarding the individual in question, does not seem sufficient to warrant a conclusion of guilt BARD. Again, suppose that the individual against whom such evidence is stacked offers an alibi that the prosecution does not fully rebut, or that there are one or more pieces of physical evidence that cast doubt on his guilt. As in the earlier homicide case, the propensity evidence does not bridge the gap between the trace evidence and a finding of guilt BARD because it is simply unknown how many other individuals might have similar propensities but have exhibited them for the first time, or who, lacking them altogether, acted out of character.

Summarizing my conclusions so far, the criminal records of individuals are unlikely to reveal much about their broader characters, though they may reveal things about their more specific propensities or past-conduct patterns. The extent to which propensity evidence is probative in a given case depends on its reliability, freshness, and specificity. It is arguably permissible to introduce highly probative propensity evidence in a criminal trial despite its tendency to bias jurors against the defendant. However, jurors should be strongly encouraged not to let their knowledge of character evidence interfere with their evaluation of the trace evidence. Even when the trace evidence against a defendant is clear and convincing, credible character evidence has more probative value in cases where the point of a trial is to determine whether a defendant acted criminally than in cases where a trial must determine if it is the defendant who acted criminally.

III. CHARACTER EVIDENCE UNDER LESS-THAN-IDEAL CONDITIONS

In my introductory remarks, I suggest that some of the debate about character evidence may derive from the failure to make a distinction between its introduction and use under ideal conditions—ones strongly conducive to error avoidance in criminal trials—and its introduction and use under less-than-ideal conditions. As we see above, fresh, reliable, and specific character evidence must be used cautiously by jurors. They must not allow it to affect their evaluation of the trace evidence and they must use it with the knowledge that both good and bad people sometimes act out of character. They must understand and accept the presumption of innocence accruing to defendants and that it is the state that bears the burden of proof in criminal cases. That burden is a stringent one, requiring the persuasive rebuttal of all evidenced alternative explanations to the state's version of events in the case. Jurors must be able to distinguish the BARD standard from the CACE and PoE standards and recognize and apply these distinctions to the various types of cases in which character evidence might be relevant. In cases where the identity of the criminal agent is at issue, they must recognize the limitations of fresh, reliable, and specific character evidence. In particular, they must keep in mind that the defendant with known criminal tendencies before them may be only one possible perpetrator of the criminal act.

Given the lengthy list of conditions that must be met for character evidence to be used appropriately and well, it is important to consider just how often these conditions are apt to be satisfied. They may, in fact, often not be, and in ways that bear directly on the advisability of introducing character evidence into criminal trials. In the first place, jurors will need considerable help from judges and defense attorneys in grasping and applying the relevant concepts and distinctions. As we see above, some advocates of relaxed character-evidence rules argue that the weakly probative kinds can be flagged by defendants' attorneys and shown to be of such dubious value that jurors (or, less often, judges) will not be misled by them.Footnote ²⁶ But not all defendants will have adequately compensated or experienced criminal-trials attorneys, and this might diminish our enthusiasm for permitting the introduction of such evidence. Still, most attorneys probably will be aware of the dangers of character evidence and motivated enough to monitor its introduction and insist that juries be given appropriate instructions in its use.

Second, it is apparent that the judges presiding over trials will vary not only in their sympathies toward defendants but also in the care with which they screen dubious character evidence or instruct jurors about its probative value. They will also vary in the extent to which they emphasize and explain clearly to jurors such crucial legal concepts as the presumption of innocence and the BARD standard (or, in the case of bench trials, understand or adhere to these requirements themselves). As Laudan makes abundantly clear in his discussion of these matters, there is widespread confusion among both judges and legal scholars about these core features of criminal trials.Footnote ²⁷ To the extent that such confusion pervades judges' decisions or their instructions to jurors about character evidence, we have reason to be concerned about how well those who try the facts will evaluate such evidence. Moreover, even if we were to imagine improved instructions from judges to jurors about these complex legal concepts and the kinds of evidence they are to consider or ignore, juror studies tell us that there is some tendency for jurors to ignore instructions from judges in cases in which these instructions conflict with what jurors regard as just outcomes.Footnote ²⁸ Defendants whose disreputable pasts are revealed might not fare well in court regardless of the strength of the trace evidence against them or the instructions given to jurors.

Third, jurors will vary widely in their levels of education, capacities for complex reasoning, and attitudes toward police, prosecutors, and the criminal justice system as a whole. They will also differ in their abilities to understand, apply, or abide by complex legal concepts. There is some evidence, for instance, that many jurors do not understand or accept the presumption of innocence, preferring instead to presume defendants guilty until they are proven innocent.Footnote ²⁹ This is an especially powerful reason to avoid putting weakly probative character evidence in their hands and it might be a reason for not providing them with highly probative character evidence. Similarly, we might wonder about how well most jurors understand the BARD, CACE, and PoE standards. Given the evident confusion among judges and legal scholars about them, it would not be surprising to find that many jurors are uncertain or confused about what they mean or how to apply them. Again, better and more frequent instructions from the judge might aid jurors in this regard. However, if reliable character evidence is introduced into a trial, judges will also have to explain the complex interplay between it and trace evidence and identify the circumstances in which character evidence is useful in enabling jurors to reach a conclusion of guilt BARD. In addition, jurors must be prepared to entertain alternative interpretations of the trace evidence. Only then will they be in a position to see how far the trace evidence carries them toward the conclusion that the defendant is guilty BARD. Yet we might wonder in general how capable or motivated jurors are to perform these complex hermeneutical tasks and in particular how capable or motivated they will be to do so if presented with evidence suggesting that a defendant has a highly unsavory character.Footnote ³⁰

Fourth, as has already been noted, our willingness to allow past-conduct evidence into criminal trials should depend in part on its reliability. This is especially true with regard to defendants' past criminal records. It is apparent that defendants' criminal records can sometimes say more about the opportunities available to them (or, in the case of severely socially deprived defendants, not available), the activities of the police (in particular, whom they tend to routinely target or investigate), or the political or personal agendas of prosecutors than they say about defendants and their characters. One does not have to be completely cynical about the police and prosecutors to recognize that they vary considerably in their capabilities, respect for the law, and integrity. Indeed, one does not have to search too far to find cases where police have done poor jobs of investigating crimes or have inexplicably targeted suspects, all the while ignoring important exculpatory evidence and other, more likely suspects, and where prosecutors have so single-mindedly focused on getting convictions that they have employed tactics (e.g., the use of jailhouse snitches) that are deeply problematic if not unethical.Footnote ³¹ One might hope that the miscarriages of justice that result from police or prosecutorial misconduct are more the exception than the rule. But their existence should make us wary of relying in an uncritical way on defendants' past criminal records. Perhaps long criminal records show something significant about defendants' character in ways that short ones do not, but even with the former we might want to exercise some caution.

Again, the preceding set of concerns has particular relevance for our evaluation of the extent or ways in which more character evidence might be used in criminal trials. Such evidence is not only potentially highly prejudicial against defendants, it is also evidence that must be utilized in ways that complicate the decision-making tasks confronting jurors. It is possible, of course, to envision reforms in the criminal justice system generally and in criminal trials specifically that would reduce the hazards of permitting jurors access to character evidence in their deliberations. More generous funding of defense counsel for indigent defendants would have salutary effects of many kinds and would likely produce more and better challenges to the introduction and use of character evidence. Better and more insistent instructions from judges in both the early and late stages of trials might also help, although, as we see above, there is evidence that jurors do not heed them in some instances.

More radically, we might consider the feasibility of replacing citizen jurors with lay judges who, together with regular judges, weigh and evaluate the evidence against criminal defendants.Footnote ³² The advantage of employing lay judges is that they gain considerable experience trying cases and might be expected to adhere better to the rules of evidence, abide by relevant legal standards of proof, and follow the instructions of the regular judges with whom they work on a continuing basis. Another possibility would be to require jurors to produce reasons for their verdicts, so that the presiding judge could better evaluate whether they have followed her instructions, observed legal standards of proof, and correctly weighed probative character evidence against trace evidence. Such a requirement will be resisted by some on theoretical or practical grounds, but it is not apparent why jurors' reasoning should be insulated from more scrutiny. Independently of such a requirement, we might strongly encourage (or better, require) judges to evaluate jury decisions more often or rigorously and set them aside when they are “unsafe,” that is, not supported by the evidence.Footnote ³³ The obligation to do this could be made especially stringent in identity cases where character evidence has been introduced alongside trace evidence, though such an obligation might exist in the nonidentity cases as well. Judges would be understood as duty-bound to determine whether the trace evidence is at least clear and convincing. Individuals convicted of crimes based in part on character evidence would have grounds for appealing their convictions in the absence of such a judicial determination.

Yet if we encourage or require judges to reject unsafe convictions, should we not also require them to overturn acquittals, especially in the nonidentity cases in which the trace evidence was, in their opinion, clear and convincing and the character evidence fresh, reliable, and specific? In other words, it might seem that I am concerned only with the correction of errors by juries when they produce inappropriate convictions of defendants, not when they produce inappropriate acquittals. But error is error, whether it is to the unjust disadvantage or advantage of defendants. This is an important point, though it raises issues that go beyond the scope of this paper. Even if we set aside the formidable constitutional obstacles to judges overturning acquittals by juries, we would have to confront the debate over the desirability of jury nullification.Footnote ³⁴ Acquittals in the face of overwhelming evidence against defendants arguably serve important purposes that convictions in spite of insufficient evidence against them do not. Hence there might be more reason to guard against the latter than the former.

Finally, as already suggested, any character evidence to be introduced against a defendant should be properly vetted by the judge, with hearings conducted away from the jury. This would permit defense counsel to challenge inaccuracies or other problematic features of the defendant's criminal record. Moreover, if the hearings were extended to character evidence that the prosecution sought to use to impeach a defendant's testimony at his trial, they would have other laudable consequences. Not only would defendants with criminal records be more likely to testify in many cases and thereby offer jurors their side of the story, but factually innocent ones with criminal records would have less to fear from doing so. As things currently stand, defendants who wish to testify risk impeachment based on their past records, regardless of the relevance of their records to the current charges against them. Such impeachment is likely to be highly prejudicial to them. Moreover, defendants who elect not to testify for fear that their past records will be raised face increased odds that they will be convicted even if they are materially innocent.Footnote ³⁵ That is not only unfortunate, it is unnecessary.

IV. CONCLUDING REMARKS

In the absence of reforms of the kinds noted in the previous section, we might reasonably continue to permit the introduction of specific and reliable past-conduct evidence, drawn from a defendant's previous convictions, that establishes an element of the prosecution's case (e.g., modus operandi) or rebuts defense claims (e.g., concerning a defendant's knowledge of how to commit a type of crime). Such evidence does not really concern the defendant's character or directly invite jurors to draw inferences from the kind of person the defendant is to a finding of guilt or innocence in the current case. Its probative value is high enough to outweigh the risk that it will have prejudicial effects. So-called character testimony by witnesses about a defendant's reputation in the community appears to have weak probative value but might be allowed since it is unlikely to have strong prejudicial effects and jurors can be cautioned against reading too much into it even by inexperienced or underpaid defense attorneys of the kinds many defendants are apt to have. The same is true for witness testimony involving direct observation of a defendant's conduct. It, too, may tell us little about the defendant's character or even specific propensities but it may provide useful evidence about a defendant's conduct in circumstances similar to those of the crime for which she is on trial. Again, even overburdened or inexperienced defense counsel should be able to challenge such evidence or warn jurors about its reliability or relevance.

However, my sense is that in the absence of substantial reforms, we should otherwise resist the introduction of a defendant's past criminal record into a trial, especially in cases where the identity of the criminal agent is at issue. Granted, a defendant's criminal record may furnish us with fresh, reliable, and specific evidence concerning her propensities. In nonidentity cases, it may be possible for jurors to resist the biasing effects of such evidence and to consider it independently of and alongside the trace evidence against defendants. There is the danger that jurors will not be able to do this and that they will not receive sufficient assistance in doing so from the judge or defense counsel for the accused. Yet these dangers seem worse in the identity cases, where, in addition, jurors must bear in mind the fact that there are likely to be other people with the relevant propensity (and some without it) who might have committed the crime in question.

It has been argued that the complications posed by character evidence are no greater in degree than or different in kind from those posed by other kinds of evidence that we routinely ask jurors to confront.Footnote ³⁶ If jurors cannot be trusted to avoid inferential error with regard to character evidence, then there is little reason to believe that they can be trusted to do so with regard to other types of evidence. But this argument seems unconvincing for two reasons. First, we must consider the real or alleged problems with each type of evidence on a case-by-case basis, both to assess the probative value of the evidence in question and to weigh the likelihood of its producing inferential errors by jurors. Second, once we have done so, we might plausibly exclude certain kinds of evidence (because their probative value is weak or nonexistent or their effects are too prejudicial) or allow them, but only if we can be reasonably assured that jurors will be cautioned about the evidence and receive appropriate help in using it. We should, in other words, resist the invitation to view or treat all kinds of evidence in the same way simply because any of them might be misunderstood or misused by those entrusted with determining the fate of criminal defendants.

Laudan argues against “defendant-friendly” rules that exclude probative evidence from trials. His view is that the presumption of innocence and the BARD standard already favor defendants (and rightly so, given our strong preference for not convicting the innocent over convicting the guilty). Rules barring the introduction of character evidence, in addition to being inconsistent, make it even more difficult for the state to convict defendants and thereby create “distributionist creep.”Footnote ³⁷ Instead of protecting the innocent, such rules tend to protect the guilty, and that is a kind of error that we should equally attempt to avoid. Yet Laudan seems more convinced of the probative value of a defendant's past criminal record than I am and more sanguine about the abilities of jurors to utilize character evidence properly. The more important point is this: in urging considerable caution in the use of character evidence, my aim is not to add protections to defendants that go beyond the presumption of innocence and the BARD standard. Rather, it is to try to ensure that the presumption of innocence is not subtly undermined by the introduction of a defendant's past criminal record and that defendants are convicted only when the trace and character evidence, taken together, satisfy the BARD standard.