An Empirical Study of Rule 609 and Suggestions for Practical Reform
An Empirical Study of Rule 609 and Suggestions for Practical Reform
Ric Simmons 0 1 2
0 Thi s Article is brought to you for free and open access by the Law Journals at Digital Commons @ Boston College Law School. It has been accepted for inclusion in Boston College Law Review by an authorized editor of Digital Commons @ Boston College Law School. For more information , please contact
1 Ric Simmons, An Empirical Study of Rule 609 and Suggestions for Practical Reform , 59 B.C.L. Rev. 993,
2 Th e Ohio State University Moritz College of Law , USA
Part of the Criminal Law Commons; Criminal Procedure Commons; and the Evidence Commons
Abstract: Rule 609 of the Federal Rulesof Evidence allows a party
tomipeach a witness with his or her prior criminal convictions. It is fair to say that
this rule is the most criticized of all the Rules of Evidence; scholars have been
calling for its reform or outright abolition for decades. These critics argue that
the rule relies on propensity evidence, which has very little probative value in
evaluating a witness’s truthfulness on the stand, and t—haetspecially when
used to impeach a criminal defendant—the evidence carries a high risk of
unfair prejudice and often prevents defendants from testifying at trial. What has
been missing from the debate so far is what is actually happening when judges
apply Rule 609 in the courtroom. ThisArticle conducts an empirical study of
Rule 609 to determine how the rule operates in practice. First, theArticle
presents a historical background of Rule 609 and evaluates some of the common
criticisms of the rule. Next, the Article presents the results of a survey of law
students and federal district court judges to determine how much probative
value and unfair prejudice each group perceives for different types of prior
convictions. The survey finds some differences between the groups, but
ovreall notes a surprising consensus that crimes of theft have a high probative
value for proving lack of credibility, whereas othre types of convictions do not.
The Article then examines how federal district court judges actually apply
Rule 609 in the courtroom. Contrary to conventional wisdom, it shows that
federal judges do not routinely admit prior convictions to impeach criminal
defendants, and that (consistent with the survey) judges tend to admit theft
crimes more often than almost any other type of conviction. The review of
district court decisions does indicate some extreme outliers and an
unexplained and troubling tendency toadmit crimes of drug possession.The
Article then proposes a modest reform to Rule609, which encourages the
admission of theft crimes to impeach witnesses, but precludes the use of other types
of criminal convictions as impeachment evidence.
© Ric Simmons, 2018. All rights reserved.
* Chief Justice Thomas J. Moyer Professor for the Administration of Justice and Rule of Law,
Michael E. Moritz College of Law, The Ohio State University. I would like thank Angela Lloyd,
Anna Roberts, Daniel McConkle, and the participants of the 2017 CrimFest conference for their
helpful suggestions and feedback. I would also like to thank Tresha Patel, Kelsey Kornblut, and
Melissa Lenz for their excellent research assistance on this Article.
Rule 609 is perhaps the most maligned of any Federal Rule of
Evidence. For decades, scholars have consistently argued for its reform or
oturight abolition.1 Specifically, the critics have targeted the application of
Rule 609 to criminal defendants; that is, the practice of impeaching a
criminal defendant with a prior conviction2. The critics argue that prior
convictions have very little probative value to prove dishonesty on the stand; they
may not be reliable indicators of actual criminal activity; the threat of their
admission deters defendants from exercising their constitutional right to
testify; and they carry a significant danger of unfair prejudice, because a
1 The scholarly critiques of Rule 609 are too numerous to list infull here. Some of them even
pre-date the rule itself. See, e.g., Richard Friedman, Character Impeachment Evidence:
PsychoBayesian [!?] Analysis and a Proposed Overhaul, 38 UCLA L. REV. 637 (1991) (proposing
abolition of Rule 609 and reform ofRule 608 to never allow any character evidence to impeach
criminal defendants); Leslie Alan Glick, Impeachment by Prior Convictions: A Critique of Rule 6-09 of
the Proposed Rules of Evidence for U.S. District Courts, 6 CRIM. L. BULL. 330 (1970) (discussing
the legal questions and implications of Rule 609, whether it is consistent witchase law, and the
fundamental policy questions involved with Rule 609 prior to its adoption); Alan D. Hornstein,
Between Rock and a Hard Place: The Right to Testify and Impeachment by Prior Conviction, 42
VILL. L. REV. 1 (1997) (arguing that prior convictions should never be allowed to impeach
criminal defendants); Gene R. Nichol, Jr., Prior Crime Impeachment of Criminal Defendants: A
Constitutional Analysis of Rule 609, 82 W. VA. L. REV. 391 (1980); Anna Roberts, Conviction by Prior
Impeachment, 96 B.U. L. REV. 1977 (2016) [hereinafter Roberts, Conviction] (reviewing multiple
reasons why Rule 609 is flawed and should be abolished); Anna Roberts, Impeachment by
Unreliable Conviction, 55 B.C. L. REV. 563 (2014) [hereinafter Roberts,Unreliable Conviction]
(discussing the extent to which convictions are reliable indicators of relative culpability); Anna
Roberts, Reclaiming the Importance of the Defendant’s Testimony: Prior Conviction Impeachment
and the Fight Against Implicit Stereotyping, 83U. CHI. L. REV. 835 (2016) [hereinafter Roberts,
Implicit Stereotyping] (arguing that Rule 609 keeps defendants off the witness stand, thus
removing an opportunity to lessen the effects of jurors’ implicit bias against minority defendants);
Robert G. Spector, Impeaching the Defendant by His Prior Convictions and the Proposed Federal
Rules of Evidence: A Half Step Forward and Three Steps Backwar1d,LOY. U. CHI. L.J. 247
(1970) (arguing for the abolition of Rule 609 because impeachment by prior conviction has little
relevancy, produces a “chilling effect” on criminal defendants, and creates an “associational
effect” with witnesses testifying on the defendant’s behal[fh)ereinafter Spector, Impeaching the
Defendant]; Robert G. Spector, Commentary, Rule 609: A Last Plea for Its Withdrawal, 32 OKLA.
L. REV. 334 (1979) [hereinafter Spector, Rule 609]. But see generally Victor Gold, Impeachment
by Conviction Evidence: Judicial Discretion and the Politics of Rul6e09, 15 CARDOZO L. REV.
2295 (1994) (arguing that the primary problem with Rule 609 is that judges do not properly apply
its balancing test and too often admit prior convictions against defendants).
2 Most instances of impeaching a criminal defendant with a prior conviction are covered by
Rule 609(a)(1)(B), but criminal defendants can also be impeached under Rule 609(a)(2) if the
prior conviction was a crime of falsity. Because many of the reform proposals argue for abolishing
the practice of impeaching acriminal defendant with any prior conviction, I will use the term
“applying Rule 609 to a criminal defendant” rather than simply referring to Rule 609(a)(1)(B).
jury will tend to ignore the limiting instruction and use the prior convictions
as evidence that the defendant has a propensity to commit a crime.3
And yet the rule endures, at least on the federal leve4l.Indeed, on the
state level there is a trend toward liberalizing the rules on impeaching with
prior convictions.5 Critics of the rule have offered a number of reasons,
some of them somewhat sinister, for why the rule persis6tsO. thers argue
that the politicians and policymakers who maintain the rule are acting in
good faith, but that they have a wildly inaccurate understanding of the
porbative value and unfair prejudice of this type of evidence.7
But there is another possible explanation for the continued perserv-e
ance of Rule 609: the critics may have it wrong, and the policymakers who
drafted the rule may have been right all along. After all, in most cases Rule
3 See Friedman, supra note 1, at 678–80 (concluding that character impeachment evidence
has no probative value for credibility and is highly prejudicial); Hornstein,supra note 1, at 9–20,
40–55 (arguing that prior convictions have little probative value and impeachment by prior
conviction burdens the defendant’s constitutional right to testify); Roberts,Conviction, supra note 1,
at 1992–2001 (addressing the flawed assumptions of Rule 609;)Roberts, Unreliable Conviction,
supra note 1, at 579–92 (discussing the unreliability of prior convictions as indicators of relative
4 For an excellent discussion of successful efforts to abolish the application of Rule 609 to
criminal defendants on the state level,see Roberts, Conviction, supra note 1, at 2018–36. There
have been two substantive amendmentsto Rule 609 on the federal level, but neither one changed
the balancing test for admitting prior convictions against criminal defendantsS. ee FED. R. EVID.
609 advisory committee’s notes.
5 See Roberts, Conviction, supra note 1, at 1990 (citing Dannye R. Holley, Judicial Anarchy:
The Admission of Convictions to Impeach: State Supreme Courts’ Interpretive Standard,s1990–
2004, 2007 MICH. ST. L. REV. 307, 315.
6 Some proposed reasons why Rule 609 has escaped reform efforts: actors in the criminal
justice system secretly want jurors making decisions based on criminal propensity evidence; they
are further concerned that any reform of the rule would open up deeper, darker questions about the
legitimacy of the convictions themselves; and they are more than happy to deter defendants from
testifying, because this will improve conviction rates and reduce the ability of a juror to empathize
with the defendant. See id. at 2015–17.
7 The legislative history surrounding Rule 609 implies that the members of Congress who
supported the rule honestly believed that a defendant’s prior convictions were relevant to a jury in
determining the party’s credibility. See, e.g., 120 CONG. REC. 37,076 (1974) (statement of Senator
McClellan in arguing that Rule 609 should automatically admit all prior felony convictions to
impeach criminal defendants):
Can it really be argued that the fact that a person has committed a serious crim—ea
felony—has no bearing on whether he would be willing to lie to a jury? Should a
jury be denied that right [to hear the evidence]? Should society be denied the
opportunity, in trying to protect itself, in its effort to discover the truth, to show that the
witness before it is a man who has committed such a crime and, therefore, might be
willing to now lie to a jury?
609 does not require a judge to admit these prior conviction8s;it merely
allows the judge to admit the evidence if she determines that the probative
value of the evidence to impeach the defendant outweighs alul nfair
prejudice. And after applying this balancing test, judges often admit these prior
convictions, though not nearly as often as critics imp.l9yThis means that
judges frequently believe that these prior convictions are useful to a jury in
evaluating a defendant’s credibility—so useful, in fact, that their usefulness
outweighs their rather substantial unfair prejudice.10
It could be true, of course, that these judges are misguide—dthat they
are miscalculating the probative value or the unfair prejudice of thisi- ev
dence, either for illegitimate reasons or simply because of an honest
misunderstanding of how useful the evidence is and/or how much it may unfairly
impact the jury.11 To help determine what is really going on when judges
make their Rule 609 detemr inations, this Article presents the results of a
survey that was given to federal trial judges to see how they measured the
probative value and unfair prejudice of various prior convictions. The same
survey was given to a number of second-year law students, to see what
differences there might be between the opinions of those who had been
appyling the rule for decades and those who had just been exposed to the rul1e2.
Perhaps unsurprisingly, the judges are far more likely to admit prior
convictions under this rule, but even the students (who show great skepticism
about the rule in general) are willing admit to prior convictions for certain
types of crimes.13 Thus, among the experts and the novices alike, there is a
consensus that an outright ban on this evidence is inappropriate; that in at
least some circumstances, admitting a prior conviction to impeach a
crimi8 Rule 609(a)(2) requires admission of prior convictions that are crimes of falsity, if the
conviction is less than ten years old. FED. R. EVID. 609(a)(2). As explained below, this Article focuses
on Rule 609(a)(1)(B), which covers the admissibility of convictions that are not crimes of falsity
when offered to impeach a criminal defendant. Id. R. 609(a)(1)(B).
9 As noted in Part III, federal judges admit prior convictions only about two-thirds of the
time, and they allow the jury to hear the name of the crime only about half of the timeS.ee infra
notes 152–164 and accompanying text (analyzing the admission of prior conviction evidence for
impeachment of criminal defendants in 120 federal district court cases).
10 Unsurprisingly, courts that allow prior convictions to be admitted to impeach criminal
defendants routinely mention the probative value of this evidence to prove lack of credibility.See,
e.g., Walden v. GeorgiaP-ac. Corp., 126 F.3d 506, 523 (3d Cir. 1997) (quoting Cummings v.
Malone, 995 F.2d 817, 826 (8th Cir. 1993) (“[O]ne who has transgressed society’s norms by
committing a felony is less likely than most to be deterred from lying under oath.”)).
11 See supra note 6 (discussing possible reasons the criminal justice system resists reform
efforts to Rule 609).
12 See infra notes 130–135 and accompanying text (describing the process for conducting the
13 See infra notes 136–138 and accompanying text (analyzing the results of the conducted
nal defendant does more good than harm.14 Of course, all of the judges and
the law students could simply be wrong—perhaps they all miscalculate the
probative value and/or the unfair prejudice of this type of evidence. But the
breadth and the strength of this consensus should at least give the Rule 609
reformers pause before they call for the outright abolition of this rule1.5
Instead, a more modest reform of Rule 609 may be in order, perhaps one that
enhances the balancing test, or that specifies which types of crimes may be
used to impeach.16
In addition to the survey results, thAerticle examines what federal
judges do in the courtroom when faced with applying Rule 609 to criminal
defendants. The Article analyzes over 120 federal district court cases in
which the prosecutor sought to admit a prior conviction to impeach the
defendant.17 The results of this analysis are broadly consistent with our r-su
vey: judges almost routinely admittedevidence of certain types of crimes,
whereas evidence of other types of crimes was not usually admitted (or at
least the jury was not told the name of the crime for whicha conviction
Studying the application of Rule 609 also has broader implications for
evidence law. Balancing test—sparticularly those that balance probative
value and unfair prejudice—are extremely common in deciding whether to
admit evidence,19 and judges are given vast amounts of discretionin
applying those balancing tests.20 But in most circumstances,their decisions are
very fact-specific; the unfair prejudice and probative value of a graphic
photograph or of a prior bad act offered to prove knowledge will vary
depending on the facts of the case and the details of the proffered evidence.
Thus, it is very difficult to conduct any systematic study about how judges
apply a Rule 40-3type balancing test in most cases. In contrast, we can
study Rule 609’s balancing test in a much more objective way: assuming
the prior conviction is for a type of crime that is completely dissimilar from
14 See infra Part II.B and accompanying text.
15 See generally, e.g., Spector, Rule 609, supra note 1 (arguing for the abolition of Rule 609).
16 Another possibility would be to codify all or part of the fivef-actor test that is used in
various forms by the circuit courts when applying Rule 609’s balancing test to criminal defendants.
See, e.g., United States v. Hernandez, 106 F.3d 737, 739–40 (7tChir. 1997) (restating the
fivefactor test first enumerated in United States v. Mahone, 537 F.2d 922 (7th Cir. 1976)).
17 See infra notes 152–164 and accompanying text (discussing the results from an analysis of
federal district court cases applying Rule 609).
18 See infra notes 152–164 and accompanying text.
19 Indeed, Rule 403, which bars evidence ifthe unfair prejudice substantially outweighs its
probative value, applies to nearly every evidentiary ruling made by a trial judge. See FED. R. EVID.
20 See, e.g., Gold, supra note 1, at 2321–27 (arguing that, with respect to Rule 609, the
problem rests with the discretion granted to judges in applying the rule).
the crime for which the defendant is now on trial2,1 the probative value for
impeachment purposes and unfair prejudice of the evidence should remain
relatively constant for any trial.22 Thus, it is possible to ask judges, lawyers,
or even lay peopleto determine the degree ofprobative value and unfair
prejudice for every type of criminal conviction, anmdake objective
decisions as to which prior convictions should be admitted and which should
not. This microcosm can give us insight into how judges apply—or should
apply—balancing tests in many other contexts, by seeing how judges
measure probative value and unfair prejudice. Furthermore, we masyee
differences between the decontextualized survey results and the actual results at
trial, where a judg’es decision may be improperly influenced by implicit
bias, the strength of the prosecutor’s case, or any other legally irrelevant
Part I of the Article reviews the history of Rule 609, and then critically
examines the four main critiques of applying the rule to criminal
defendants.23 Part II describes the survey and then presents and analyzes thee- r
sults.24 Part III compares the results of the survey with how trial judges are
actually deciding Rule 609 motions.25 Part IV of this Article proposes some
modest changes to Rule 609 in response to this research.26
I. RULE 609
A. History of the Rule
Historically, under the common law, a prior criminal covniction could
preclude a witness from testifying altogeth2e7r.By the time the Federal
21 If the prior conviction is for a crime that is similar to the crime for which the defendant is
now on trial, the likelihood of unfair prejudice increases substantially, as there is a greater danger
that the jury will use the prior crime for an illegitimate propensity purposeS.ee, e.g., Hernandez,
106 F.3d at 740 (“The court was well aware that there was a similarity between the two crimes, a
factor that requires caution on the part of the district court to avoid the possibility of the jury’s
inferring guilt on a ground not permissible under Rule 404(b).”).
22 Of course, the probative value and unfair prejudice of a given prior conviction cannot be
held completely constant in real-life cases; for example, juries in certain parts of thecountry may
see gun possession crimes as more or less prejudicial, depending on the local cultural attitudes
towards guns. And the trial judge is also supposed to consider the “importance of the defendant’s
testimony” in the case, which may vary depending on the crime for which he is being charged and
his theory of defense. See id.
23 See infra notes 27–129 and accompanying text.
24 See infra notes 130–151 and accompanying text.
25 See infra notes 152–164 and accompanying text.
26 See infra notes 165–180 and accompanying text.
27 See Jeffrey Bellin, Circumventing Congress: How the Federal Courts Opened the Door to
Impeaching Criminal Defendants with Prior Convictions, 42 U.C. DAVIS L. REV. 289, 296 (2008);
Christian A. Bourgeacq, Note,Impeachment with Prior Convictions Under Federal Rule of
EviRules of Evidence were codified, almost all American jurisdictions had
abolished this absolute bar2,8 but all still allowed prior convictions to
impeach a witness, under the theory that “[t]here is little dissent from the
general proposition that at least some crimes are relevant to credibility2.”9 The
Advisory Committee’s original version of Rule 609 automatically admitted
all felony convictions as well as all crimes of falsity, with no discretion
gvien to the trial judge.30 When a subcommittee of the House Judiciary
Committee considered the Rules of Evidence, many of the witnesses argued that
the trial judge should have discretion to preclude prior convictions if the
unfair prejudice of the evidence substantially outweighed its probative
vlaue.31 The subcommittee added this balancing te—st essentially stating that
Rule 403 should apply to prior conviction impeachment evidenc—efor
felony convictions that were not crimes of falsit3y2. The full Judiciary
Committee went even further and precluded all convictions that were not crimes
of falsity, and, after an extensive floor debate, this version passinedthe
House.33 The Senate passed a version of Rule 609 that was similar to the
Advisory Committee’s initial version, automatically admitting all felonies
and prior crimes of falsity.34 The Conference Committee thus created a
compromise between the two bills: automatically admitting all crimes of falsity
but applying a revised Rule 403 balancing test to other felonies, which would
only be admitted if the probative value of the evidence outweighed itsn-u
dence 609(a)(1): A Plea for Balance, 63WASH. U. L.Q. 469, 470 (1985) (citing C. MCCORMICK,
MCCORMICK ON EVIDENCE 93 (3d ed. 1984)). This prohibition was erased gradually over the end
of the 19th Century and the beginning of the20th Century as statutes were passed that permitted
people with felony convictions to testify.See Montgomery v. United States, 403 F.2d 605, 611
(8th Cir. 1968) (discussing the common law prohibition on individuals testifying if they had a
criminal record and its abolition in the United States during the nineteenth century).
28 FED. R. EVID. 601 advisory committee’s note. The blanket ban on felons testifying as wt-i
nesses was removed from all state laws by 1953. 3 JACK B. WEINSTEIN & MARGARET A. BERGER,
WEINSTEIN’S FEDERAL EVIDENCE § 601 app. at 01 (Mark S. Brodin & Joseph M. McLaughlin
eds., 2d ed. 2016).
29 FED. R. EVID. 609 advisory committee’s note.
30 See Gold, supra note 1, at 2298–301 (discussing the history of Rule 609(a)). The Advisory
Committee briefly changed its mind and added a Rule 403 balancing test to thruele, but
subsequently changed it back before submitting the Rules of Evidence to Congress in 1972. Id.
31 Id. at 2301.
32 Id. at 2301–02.
33 Id. at 2303 & nn.45–46. Professor Goldnotes that the debate on Rule 609 “far exceeded
that relating to any other provision in all the proposed Federal Rules of Evidence.” Id. at 2303.
34 Id. at 2305–07 & nn.57–60.
35 Id. at 2307–08.
The Advisory Committee substantively amended Rule 609 in 1990 and
2006.36 The 1990 amendment relaxed the balancing test for criminal
convictions used to impeach any witnessother than a criminal defendant; such a
conviction would now be admissible unless the opposing party could
demonstrate that its unfair prejudice substantially outweighed its probative value.37
Admitting a prior conviction to impeach a defendant, however, still faced
the same strict balancing test: a judge is only permitted to admit such a
conviction if the prosecutor could demonstrate that the probative valuteo
impeach outweighs its unfair prejudice38. The Advisory Committee left this
balancing test in place because:
[T]he rule recognizes that, in virtually every case in which prior
convictions are used to impeach the testifying defendant, the
defendant faces a unique risk of prejudi—ce i.e., the danger that
convictions that would be excluded under Federal Rule of
Evidence 404 will be misused by a jury as propensity evidencee-d
spite their introduction solely for impeachment purposes.39
The 2006 amendment limited the number of crimes that would be
automatically admissible as “crimes of falsity.”40
The history of Rule 609 suggests a compromise between those who
believe that prior convictions shouldnever be admitted to impeach criminal
defendants and those who argue that prior convictions shouldautomatically
be admitted to impeach.41 It also shows that the drafters of the rule were
aware of the unique danger that criminal defendants faced if their prior
convictions were admitted against them by creating an unusually high barrier to
admissibility in such circumstances.42 The unfair prejudice of prior
convci36 Like nearly every other Federal Rule of Evidence, Rule 609 was also amended in 2011 as
part of the Restyling project, but no substantive changes were made.See FED. R. EVID. 609
advisory committee’s notes.
38 See FED. R. EVID. 609(b)(1).
39 Id. R. 609 advisory committee’s note to 1990 amendmentT.he 1990 amendment also
allowed a party to “remove the sting” by eliciting evidence of the witnsespsr’ior conviction on
direct examination, rather than waiting for the opposing party to confront the witneosns
40 See id. advisory committee’s note to 2006 amendment.After 2006, only convictions that
included a false statement or false action as an element of the crime would be classified as a crime
of falsity. Id.
41 See Gold, supra note 1, at 2298–309 (recounting the history of Rule 609 and its adoption;)
Hornstein, supra note 1, at 6–8 (discussing the debate surrounding the adoption of Rule 609).
42 See FED. R. EVID. 609 advisory committee’s note; Gold, supra note 1, at 2298–309;
Hornstein, supra note 1, at 6–8. The standard balancing test for admissibility, found in Rule 403,
strongly favors admissibility: the party opposing the evidence must prove to the judge that the
danger of unfair prejudice substantially outweighs its probative value. There are only three
provition evidence is already much higher for criminal defendants;43 thus, even if
the standard Rule 403 balancing test applied in this context (as it does for
impeaching any other witness with a prior conviction), very few of these
prior convictions would be admitted. The fact that the burden shifts to the
prosecutor to prove a greater probative value than unfair prejudice means
that only the most probative of these prior convictions should be admitted.44
As we will see from the survey below, most judges follow this principle in
the decontextualized setting of the survey, though they seem willing to
damit a somewhat wider variety of crimes in the courtroom setting45. And as
will always be true with a rule involving judicial discretion, there are judges
who are outliers on either side.
B. Critiques of the Rule
As noted above, there is no shortage of articles advocating reform or
repeal of Rule 609.46 Most of these critiques (and the focus of thisArticle)
have to do with a very specific provision of Rule 609: impeaching a cri
minal defendant with a prior conviction.47 But any evaluation of this aspect of
the rule must also consider the entire infrastructure of impeachment with
prior dishonest actions, including both how Rule 609 should apply to tw- i
nesses who are not criminal defendants as well as how Rule 608 regulates
the admissibility of actions that are not criminal convictio4n8s.Thus, any
proposed reforms to how Rule 609 applies to criminal defendant—sas well
as any arguments for those reforms—must take into account how those
ersions in the Rules that apply a different (and stricter) balancing test: Rule 609(a)(1)(B), when a
prosecutor attempts to admit a prior conviction to impeach a criminal defendant; Rule 703, when
an expert is asked to reveal her (otherwise inadmissible) underlying data; and Rule 609(b), when a
party seeks to impeach with a prior conviction that is over ten years old.
43 See infra notes 99–129 and accompanying text (arguing that the danger of jurors
improperly using prior conviction evidence as propensity evidence is heightened for criminal defendants).
44 See FED. R. EVID. 609 advisory committee’s note to 1990amendment (“Although the rule
does not forbid all use of convictions to impeach a defendant, it requires that the government show
that the probative value of convictions as impeachment evidence outweighs their prejudicial
45 Compare infra notes 136–138 (discussing the results of the survey),with infra notes 141–
151 (discussing the results of the analysis of the federal district court cases).
46 See supra note 1 (presenting a list of critics of Rule 609).
47 See, e.g., Hornstein, supra note 1, at 55–61 (arguing that Rule 609 may unconstitutionally
burden a criminal defendant’s right to testify); SpectorI,mpeaching the Defendant, supra note 1, at
247–51 (arguing that impeaching a criminal defendant with felonies has little relevancy, produces a
“chilling effect,” and produces an “associational effect”). This mostly includResule 609(a)(1)(B),
which covers impeaching a criminal defendant with felonies that are not crimes of falsity, but would
also include Rule 609(a)(2), which covers impeachingany witness (including a criminal defendant)
with a crime of falsity. See FED. R. EVID. 609.
48 See FED. R. EVID. 608–609.
forms and arguments would affect the rest of the impeachment provisions in
Rules 608 and 609.
The arguments against the use of prior convictions to impeach can be
roughly broken down into four categories. Some of these arguments are
specific to the practice of admitting prior convictions evidence against
criminal defendants; others are broader attacks on admitting prior convoicntsi
against any witness; some are so broad that they argue against admitting
any prior dishonest action to impeach any witness.49 The four categories are
(1) Prior convictions have little or no probative value regarding
(2) Admitting prior convictions infringes on the defendant’s right
(3) Some of these convictions are unreliable (i.e., the defendant
may not have actually been guilty of the crime); and
(4) Jurors will use the prior convictions for an improper
This Article will consider each of these criticisms in turn.
1. Probative Value of Prior Convictions for Determining Credibility
This critique appears in many different forms. The “strong” version of
this argument claims that there is no probative value whatsoever in
admitting any prior conviction5s1. Most lay peopl—eand certainly almost all
49 See Roberts, Conviction, supra note 1, at 1992–2004 (discussing the underlying, flawed
assumptions of Rule 609);Roberts, Unreliable Conviction, supra note 1, at 574–79 (discussing
common criticisms of Rule 609).
50 Professor Anna Roberts has proposed another category of critiques regarding Rule 609: that
when it is used against criminal defendants, it contributes to a “broader dysfunction” in the
criminal justice system. She gives a number of examples: it discourages criminal defendants from
taking the witness stand; it may contribute to wrongful convictions; it is a “hidden” collateral
conesquence of a criminal conviction; it exacerbates the racial disparities in the criminal justice system;
it gives even more power to prosecutors; and it influences more defendants to plea bargain their
cases. Roberts, Conviction, supra note 1, at 2004–14. I address thefirst of these critiques infra
notes 76–84 and accompanying text. Some of the other critiques include normative assumptions—
for example, a presumption that it is detrimental to give prosecutors more power, or that it is
detrimental to the system to have more plea bargains—that are beyond the scope of this paper.
51 Critics of Rule 609 point out that the rule relies on a long chain of assumptions and
inferences: (1) there is such a thing as “character for truthfulness”; (2) knowing a witness’s character
for truthfulness is useful to jurors in evaluating the witness’ credibility; (3a) witness’s character
for truthfulness can be determined by learning about the witness’s prior dishonest actions; (a4)
witness’s character for truthfulness can be determined by learning about the witness’s prior
convictions; and (5) jurors are able (with the help of a limiting instruction) to use this evidence
appropriately to evaluate a witness’s credibility. See Robert D. Dodson, What Went Wrong with Federal
judges—would reject this absolutist argument. Surely some criminal nc-o
victions—such as crimes of falsity—are relevant in determining whether a
witness is currently telling the truth. As an extreme example, if the jury
hears from two witnesses, and one has never been convicted of any crime,
whereas the other has been convicted of perjury three times in the past five
years, it would be reasonable for a juror to find the mfoerr more credible
than the latter.
But most critics of Rule 609 do not make such an extreme argument.
Instead, they argue that the probative value of prior convictions is very low,
such that it will almost always be outweighed by the unfairly prejudicial
effect that they will have on the jury (as described by the fourth objection in
Section 4 below).52 Professor Anna Roberts, who has written extensively on
this issue, argues that the justifications given by the defenders of Rule 609
“rest not on data but onwhat one might call ‘junk science at its worst.’5”3
She notes that the probative value of a prior conviction rests on a number of
(1) The defendant committed the crime in question;
(2) Those without such a conviction didnot commit the crime in
(3) The conviction can be related to a particular character trait;
(4) The defendant still possesses that trait just as he had it then;
(5) The trait helps predict the likelihood that the defendant will lie
while on the stand; and
(6) The jury will be able to use this evidence to help them assess
Rule of Evidence 609: A Look at How Jurors Really Misuse Prior Conviction Ev,id4e8nce
DRAKE L. REV. 1, 31–32 (1999); Roberts, Conviction, supra note 1, at 1992–93. As one
commentator noted (though with a slightly different chain of inferences):
The probativity of the evidence of the prior conviction canbe no stronger than the
weakest link in that inferential chain. Indeed, the probative value of the evidence of
prior conviction is the product of the probabilities of each inference necessary to
support the conclusion, and that product is perforce lower than the lowest probabl-i
ity of each of the several inferences to be drawn.
Hornstein, supra note 1, at 14 (citing Richard O. Lempert, Modeling Relevance, 75 MICH. L. REV.
52 See Hornstein, supra note 1, at 15–19 (discussing the marginal probative argument against
prior conviction impeachment evidence);Spector, Impeaching the Defendan,t supra note 1, at
Professor Roberts attacks all of these presumptions and points out that
the conclusion from a chain of inferences is suspect or invalid if only one of
those inferences is faulty.55 But most of these presumptions are in fact quite
reasonable. The correlation between conviction of a crime and actual guilt
for that crime is very high—at least 96%.56 The second presumption is true
for any impeachment evidence given to a jury. For example, in a given trial,
both witness A and witness B may have made multiple prior inconsistent
statements, but the opposing counsel is only aware of those made by tW-i
ness A. Should the judge bar the attorney from impeaching Witness A with
the prior inconsistent statements, simply because there is a chancethat
Witness B also has made undiscovered prior inconsistent statements? The
fourth presumption—whether there are fixed traits (such as “propensity to
lie”) that can be assigned to witnesse—schallenges not just Rule 609 but a
large subset of evidentiary rules, such as Rule 608 and Rule 404(a5)7. And
to the extent that fixed traits may diminish with time, Rule 609’s balancing
test already takes this into account;58 the older the conviction, the less likely
a judge will admit it, and if the conviction is over ten years old, it faces a
very strong presumption against admissibility.59
The third, fifth, and sixth presumptions all focus on the probative value
of the prior conviction to prove lack of credibility, which almost certainly
varies depending on the crime. Some scholars believe that any prior
conviction, regardless of the type of crime, indicates little or nothing abotua
defendant’s likelihood to lie on the stand.60 These critics point out that the
entire logic of the rule rests on a propensity inference (that a person who
committed a certain act in the past has a tendency to commit that act, and is
therefore more likely to commit the act in the future) and propensity
evidence is heavily disfavored in the law of evidence.61 But although
propensi55 Id. at 1992–97 (critiquing the assumptions underlying the probative value of prior
convcition impeachment evidence).
56 See infra notes 99–129 and accompanying text (discussing studies that conclude that there
is a high correlation between prior conviction and guilt).
57 See FED. R. EVID. 404(a), 608.
58 See Hernandez, 106 F.3d at 739–40 (setting out five factors for determining whether a prior
conviction is admissible against a criminal defendant under Rule 609(a)(1)(B), including the mti
ing of the prior conviction).
59 See FED. R. EVID. 609(b)(1) (admitting convictions over ten years old only if the probative
value for impeachment “substantially outweighs” the unfair prejudice).
60 See Glick, supra note 1, at 331–34 (arguing that there is no positive relationship between
the defendant’s past conviction andhis present willingness to tell the truth); Spector,Impeaching
the Defendant, supra note 1, at 249–50 (concluding that there is no logical connection between a
defendant’s past conviction and his willingness to tell the truth when testifying).
61 See, e.g., FED. R. EVID. 404(a) (barring almost any use of propensityevidence if offered to
prove actions in accordance with that propensity).
ty evidence is disfavored, it is notbanned; we allow defendants to admit
propensity evidence if they wish (and we allow prosecutors to respond); we
also allow both parties to admit evidence of prior dishonest conduct to show
that a witness has a propensity to li6e2. Thus, the drafters of the Rules of
Evidence acknowledge that propensity evidence has some probativel-va
ue.63 As we will see below,64 a majority of law students and federal judges
agree that at least some categories of criminal convictions (such as theft
crimes) have a probative value to prove propensity to lie, and are thus
uesful to prove lack of credibility on the stand.
A more sophisticated version of the argument attacking thperobative
value of prior convictions emphasizes themarginal probative value of a
criminal defendant’s prior conviction65. This tactical shift from absolute
probative value to marginal probative value is far more powerful if the
wtiness is a party to the case, and even more powerful if the witness is a
criminal defendant.66 In those scenarios, the jurors will already be aware that the
witness has a strong motivation to lie, so that their learning abtoau prior
criminal conviction will do very little to affect their judgment about the
witness’s credibility. Professor Richard Friedman imagines a juror having
the following absurd internal dialogue: “At first I thought it was veryn-u
likely that, if Defoe committed robbery, he would be willing to lie about it.
But now that I know he committed forgery a year before, that possibility
seems substantially more likely.”67
The strength of this critique rests at least in part on the typ—e and the
number—of prior criminal convictions that are availablefor impeachment
purposes. Here, it is important to consider Rule 609 not just in isolation, but
as part of the entire impeachment-by-character regime that is created by the
Rules of Evidence. We allow all witnesses to bimepeached by character
63 See id. R. 404 advisory committee’s note (discussing the rationale for admission of
64 See infra notes 136–151 and accompanying text (analyzing the data gained froma survey
of federal district court cases).
65 See, e.g., Hornstein, supra note 1, at 18 (“The question we must ask about the probative
value of prior convictions, then, is how much such evidence adds to our assessment of credibility
in light of the defendant/witness’s strong interest in the outcome.”); RobertCs,onviction, supra
note 1, at 1997 (“[T]he jury already has every reason to suspect that a defendant faced with the
loss of liberty and perhaps life might shape his or her testimony in order to maximize the
possibliity of acquittal.”).
66 See Hornstein, supra note 1, at 15–19 (discussing the marginal probative value argument).
67 Friedman, supra note 1, at 664. Professor Friedman uses a Bayesian analysis to argue that
the marginal probative value of prior convictions to impeach criminal defendants is extremely
small. See id. at 655–70 (applying a Bayesian analysis to character impeachment of criminal
witnesses who testify that the witness has a propensity to be untruth6f8ul.
And we allow all witnesses to be cross-examined with prior dishonestc-a
tions, as long as the probative value for impeachment is not substantially
outweighed by the unfair prejudice.69 If we accept the premise behind Rule
608, it would be odd to reject Rule 609 for all criminal convictions, because
in some cases, the underlying conduct that supported the criminal concv-i
tion would be admissible under Rule 608. Certainly this is true for “crimes
of falsity,” which are automatically admissible under Rule 60970 and would
almost surely pass a Rule 403 balancing test if offered under Rule 67018.
But it is also true for many crimes that do not meet Rule 609’s narrow
deifnition of crime of falsity, such as theft crime72s.For example, if a cros-s
examining attorney has a good faith belief that the witness had stolen
money from his employer, Rule 608(b) allows the attorney to ask the witness
about the prior theft to prove to hte jury that the witness had committed a
severely dishonest act in the past7.3 If the defendant had in fact
beenconvicted of stealing money from his employer, the probative value increases,
because we are more certain that the dishonest action occurred (instead of a
“good faith basis,” we have a guilty plea or proof beyond a reasonable
doubt).74 It is true that the unfair prejudice also increases, because every
criminal conviction carries a stigma; we will consider this in subsection
68 See FED. R. EVID. 608(a).
69 See id. R. 608(b).
70 See id. R. 609(a)(2).
71 Though even crimes of falsity would not be admissible under Rule 608 if the prior
convcition were identical to the crime for which the defendant is now being charged. I discuss this
problem in Part IV. See infra notes 165–180 and accompanying text.
72 See H.R. REP. NO. 93-1597, at 9 (1974) (Conf. Rep.) (“By the phrase ‘dishonesty and false
statement’ the Conference means crimes such as perjury or subornation of perjury, false statement,
criminal fraud, embezzlement, or false pretense, or any other offense in the nature ofcrimen falsi,
the commission of which involves some element of deceit, untruthfulness, or falsification bearing
on the accused’s propensity to testify truthfully.”).
73 See FED. R. EVID. 608(b).
74 But see infra notes 85–98 and accompanying text (describing how some critics dispute the
reliability of prior convictions).
75 See infra notes 99–129 and accompanying text (discussing the improper use of prior
conviction evidence by jurors as character evidence).
2. Admitting Prior Convictions Infringes on the Defendant’s Right to
The Supreme Court has held that every criminal defendant has a right
to testify in his or her own defense7.6 Furthermore, a defendant’s ability to
testify can have serious practical implications for the outcome of the case.
Studies have shown that many jurors disregard the standard instruction to
draw no inference from the defendant’s failure to testify, instead associating
the defendant’s silence with guilt7.7 Meanwhile, the threat of impeachment
by prior conviction has been proven to dissuade some criminal defendants
from testifying,78 and empirical evidence suggests that a jury is more likely
to convict a defendant if the defendant does not testif7y9. Impeachment by
prior conviction may also have played a significant role in the trials of those
who have been wrongfully convicted.80
Furthermore, as Professor Roberts has noted, if a defendant is a mme
ber of a minority group, many jurors will carry an implicit bias against the
defendant.81 If the defendant chooses to testify, he can alleviate the power
of that implicit bias, but if the possibility of impeachment by prior concv-i
tion discourages him from testifying, the jury is more likely to evaluate the
76 See Rock v. Arkansas, 483 U.S. 44, 49 (1987)(“At this point in the development of our
adversary system, it cannot be doubted that a defendant in a criminal case has the right to take the
witness stand and to testify in his or her own defense.”).
77 See LEWIS MAYERS, SHALL WE AMEND THE FIFTH AMENDMENT? 21 (1959) (survey
shows that 71% of respondents inferred guilt from a defendant’s refusal to testiJfoyh);n H.
Blume, The Dilemma of the Criminal Defendant with a Prior Record—Lessons from the
Wrongfully Convicted, 5 J. EMPIRICAL LEGAL STUD. 477, 478 & n.1 (2008) (citing a surveyfinding that
approximately half of Americans believe that refusal to testify is an indicator of g)u;ilstee also
Friedman, supra note 1, at 667 (jurors tend to ignore the judicial instruction not to infer guilt from
silence because the instruction is “virtually incoherent”).
78 According to one study, 62% of defendants without criminal records testified, but only
45% of defendants with criminal records testified. Blume, supra note 77, at 490 n.49.
79 See Jeffrey Bellin, The Silence Penalty, 103 IOWA L. REV. 395, 413 (2018) (surveying four
hundred mock jurors and finding that based on the same fact pattern, a defendant who testifie—s
and is not impeached with any prior conviction—s is convicted 62% of the time, whereas a
defendant who does not testify is convicted 76% of the time); Robert D. Okun, Character and
Credibility: A Proposal to Realign Federal Rules of Evidence 608 and 609, 37 VILL. L. REV. 533, 554–
55 (1992) (discussing scientific studies that conclude a jury is more likely to convict a defendant
who does not testify).
80 Blume, supra note 77, at 479 (“Virtually all of the [wrongfully convicted] defendants who
did not testify had a prior record that likely would have been disclosed to the jury had they taken
the stand.”). Professor Blume conducted an empirical study of 172 individuals who had been
exonerated by the Innocence Project through the end of 2006. Id. at 488. He found that of the
wrongfully convicted defendants, 91% of those who had criminal records waived their right to testify. Id.
81 See Roberts, Implicit Stereotyping, supra note 1, at 860–69 (discussing the implicit racial
stereotypes that arise when minority defendants do not testify).
defendant as a member of a (disfavored) group rather than as an indivui-d
al.82 Thus, if a black or a Latino defendant is discouraged from testifying
because of this rule, his decision could exacerbate pre-existing racial biases
in our jury system.
These critiques are powerful—but they rest entirely on the
presupmtion that using prior convictions to impeach is improper in the first place.
After all, there are many legitimate ways of impeaching a criminal
defednant: showing that he is biased; confronting him with prior inconsistent
statements; using negative character witnesses to provethat he is not
credible; or even using cross-examination to expose his story as not credible and
full of inconsistencies.83 The threat of any of these impeachment techniques
could conceivably convince a defendant that he is better off not taking the
stand, and in this way “infringe” on his right to testify8.4 Nobody would
argue, however, that allowing any form of impeachment violates the
defendant’s right to testify; surely the prosecutor has the right to conduct an
efective cross-examination. Thus, this critique of Rule 609 is only persuasive if
you have already established that admitting prior convictions to impeach is
unfair to the defendant. In other words, if you have already established that
Rule 609 evidence is of low probative value and/or allows substantial unfair
prejudice, the argument that the rule discourages the defendant from
testifying demonstrates that the damage done by the rule can be extensive. But the
argument does not, on its own, prove that the rule should not exist.
3. Some of the Prior Convictions May Be Unreliable
A number of critics of Rule 609 have attacked threule based on the
theory that the prior conviction itself may have been inaccurate—that is, the
defendant may have been wrongfully convicted of the prior crime, and thus
providing the jury with this evidence may mislead the ju85ryT.he critics
also argue that the disparity of resources between the prosecutor and the
82 See id. at 860–82 (discussing implicit stereotypes and proposing that courts focus on the
Mahone factor of “importance of the defendant’s testimony” in order to preclude evidence of prior
convictions and allow minority defendants to testify).
83 See Friedman, supra note 1, at 669 & n.80 (discussing other factors arising from
impeachment evidence that may dissuade a defendant from testifying).
84 See id.; Nichol, supra note 1, at 400–05 (discussing the burden prior conviction
impeachment evidence places on a defendant’s constitutional right to testify).
85 See Hornstein, supra note 1, at 9–12 (arguing that the prevalence of plea bargains and the
coercive nature of many plea bargains means that often a conviction is not a reliable indicator that
the defendant committed the crime); Roberts, Unreliable Conviction, supra note 1, 579–80
(arguing that courts should undertake investigations into the reliability of prior convictions prior to
admitting them under Rule 609).
defense attorney means that a criminal trial is frequently not a “fair fight.”86
Furthermore, they point out that most criminal convictions are the result of
plea bargaining, in which the prosecutor wields an extraordinary amount of
power and the defendant may be coerced into pleading guilty.87
There are three responses to this argument. First, it is again important
to compare Rule 609 to Rule 608, which allows any witness tom-be i
peached with any instance of prior dishonest conduct, as long as the
attorney has a “good faith basis” that the witness committed the dishonest
cnoduct.88 This is obviously a much lower standard of proof than what ies- r
quired to obtain a criminal conviction, however flawed one may believe our
criminal justice system to be. In the case of a prior conviction, we know that
the witness either admitted to the action under oath during a plea allocution
or he was found guilty beyond a reasonable doubt by a jury of his peers.
Indeed, one of the reasons why criminal convictions are given their own
rule separate and apart from Rule 608 is because—unlike a mildly
substantiated allegation of dishonest conduct—we can be relatively certain that the
witness committed the underlying criminal action.89
A second response is that the unreliable conviction argument is based
on a very radical proposition that involves a sweepingand widespread
indictment of our criminal justice system. Individuals are imprisoned, fined,
86 See Roberts, Unreliable Conviction, supra note 1, at 582.
87 Id. at 580–84 (concluding that it would be a “miracle” if the result of the plea-bargaining
process was reliable).
88 See FED. R. EVID. 608(b). Professor Roberts further argues that some of these felony
convictions are malum prohibitum rather than malum in se; thus, the individuals who are convicted of
these crimes are not committing a moral wrong and the prior convictions have evenless probative
value to prove lack of credibility.See Roberts, Unreliable Conviction, supra note 1, at 587–90.
This is a good point, but it merely means that notall felonies should be admitted under Rule 609;
in other words, judges should take the type of crime under consideration when using Rule 609’s
89 One possible counterargument to this response is that Rule 609 evidence cairers with it a
special (and perhaps inaccurate) veneer of reliability. In other words, when jurors hear about prior
dishonest conduct under Rule 608(b), they will automatically (and correctly) discount the strength
of the evidence because they know there is a chance that the prior dishonest conduct did not occur.
In contrast, when jurors hear about a prior conviction under Rule 609, they incorrectly assume that
it is certain that the defendant committed thecrime, and therefore fail to discount the evidenceto
account for the possibility that the defendant may not, in fact, have committed ctrhimee. The
strength of this critique rests on two factors. First, how reliable in fact are prior convictions—if the
false conviction rate is really only around one or two percent, the difference between the jurors’
belief in the defendant’s guilt and the likelihood that he is actually guilty is negligiblSe.ee infra
notes 95–98 and accompanying text (discussing how empirical evidence points to a high
correlation between conviction and guilt). Second, how much do jurors in fact believe that a prior
conviction means actual guilt? The high-profile cases involving exonerations have given many potential
jurors a strong level of skepticism about the reliability of our criminal justice system; many
potential jurors may incorrectly believe that the false conviction rate is in fact much higher than it is.
and suffer the stigma of a criminal record based on a criminal conviction. If
we do not believe criminal convictions are reliable, there are hundreds of
other rules in the criminal justice system that need to be reformed to reflect
that fact—from sentencing rules for recidivists to civil forfeiture laws to the
imposition of dozens of different collateral consequences. And most
importantly of all, we need to be fundamentally reforming the criminal justice
system itself to lower the number of wrongful convictions.90 In other words,
if criminal convictions are as unreliable as these critics contend, the
application of Rule 609 to criminal defendants is the least of our concerns.
The third response springs naturally from the previous response. Given
the widespread acceptance—both within the criminal justice system and
among the public at large—that a person convicted of a crime is almost
certainly guilty of that crime, theburden of persuasion naturally rests on those
who would argue against this proposition.91 To put it simply, the critics have
failed to make their case. The critics present two primary arguments. First,
they argue that the adversary system is unfair; the prosecutor has so many
more resources and so much more power than the defendant that the
defendant is at a disadvantage both at trial and during the plea bargaining
stage.92 This argument fails to consider the significant procedural proc-te
tions given to defendants, includingthe prosecutor’s need to prove guilt
beyond a reasonable doubt; the defendant’s right—unique among all parties
in the adversary system—to remain silent both before and during the trial;
and discovery rules, which nearly always favor the defendant. At any rate,
imbalances in resources on the one side or favorable procedural rules on the
other side are only indirectly relevant to the actual question: how often is a
defendant convicted of a felony that he in fact did not commit?
Once again, the burden of proof should rest on those challenging the
conventional wisdom on this question. The critics do indeed provide some
empirical data, such as the increasing number of exonerations of those who
had been found guilty, including those who have pledguilty.93 This is
presented as evidence that “it cannot be taken as a given that a conviction
cro90 See, e.g., Roberts, Unreliable Conviction, supra note 1, at 579–90 (discussing systemic
issues with the criminal justice system that raise questions as to the reliability of convictions).
91 See infra notes 95–98 and accompanying text (discussing empirical evidence poinitng to a
high correlation between conviction and guilt).
92 See Roberts, Unreliable Conviction, supra note 1, at 579–80 (discussing disparities in the
judicial process and the lack of a fair fight between prosecutors and defense attorneys).
93 For example, Professor Roberts points out that as of 2014 theNational Registry of
Exonerations contained 1,339 exonerees, approximately 10% of whom pled guilty. Roberts,Unreliable
Conviction, supra note 1, at 584 (citingExonerations, NAT’L REGISTRY OF EXONERATIONS, http://
For each of these crimes, the perceived probative value to proven-u
truthful character is very similar between the two groups, but the unfair
prejudice is significantly different, with judges believing the unfair effect on
the jury is less than students believe. (Note that this is not true for all the
crimes; the average unfair prejudice rating for all crimes for students was
56.2, whereas the average unfair prejudice rating for all crimes for judges
was 53.5).146 Thus, for these specific crimes only, judges appear to trust
juries more with this kind of information.
On the other hand, compare the following:
This was by far the biggest difference in favor of the students about
probative value; more students than judges seem to genuinely believe that a
conviction involving illegal immigration is useful in evaluating a witness’s
145 This Table is permanently available ahtttp://www.bc.edu/content/dam/bc1/schools/law/
146 See supra note 146 and accompanying text (analyzing the average unfair prejudice rating
for all crimes in Table 6).
147 This Table is permanently available ahtttp://www.bc.edu/content/dam/bc1/schools/law/
credibility. This was also the only crime that more students would admit
than judges (although still only 28% of students would admit the crime). On
average, students gave this crime a probative value rating approximately as
strong as many of the theft crimes, such as carjiancgk, grand theft auto,
robbery, and shoplifting.
D. Critiques of the Survey
The survey results can be critiqued on a number of levels. Mostnf-u
damentally, they do not necessarily reflect thaectual probative value and
actual unfair prejudice of each of these pieces of evidence; rather, they
represent the students’ and judgese’stimates of probative value and unfair
prejudice. Nevertheless, the survey results may be the best estimates that we
can make. It would be very hard, if not impossible, to measure tahcetual
probative value or the actual unfair prejudice of each type of prior concv-i
tions.148 The judges’ estimates may be the more accurate of the tw—o after
all, judges estimate probative value and unfair prejudice on a daily basis for
all sorts of evidence; if we ignore their estimates in this context, we are
calling into question thousands of Rule 403 rulings that judges make in cotu-r
rooms every day. The students’ responses may be less reliable indicators in
some ways, because students do not have the experience and expertise that
the judges do. The students, however, may reflect a perspective that is less
hardened by having presided over years of criminal trials and therefore
more accurate.149 They also may present a more politically diverse group of
respondents, because judges tend to be more conservative and are often
One indicator of the accuracy of the results, however, is the
consistency seen between the two diverse groups of respondents. Both groups would
admit serious theft crimes such as embezzlement and insider trading, and
148 Probative value in this context would measure how much a prior conviction in fact
indicates that a witness is likely to lie on the stand; unfair prejudice would measure how likely a juror
is to ignore the limiting instruction, and if so, to what extent the average juror will allow the belief
in the defendant’s propensity to commit a crime to influence her verdict.
149 Law students are also probably better respondents than the general population because
they are familiar with the terms “probative value” and “unfair prejudice.”
150 See Adam Liptak, Why Judges Tilt to the Right, N.Y. TIMES (Jan. 31, 2015), https://www.
[https:p/e/rma.cc/6PWXY7YA] (discussing a study that shows judges tend to be more conservative than lawyers and
possible reasons why this trend occurs)S.ee generally Adam Bonica & Maya Sen,The Politics of
Selecting the Bench from the Bar: The Legal Profession and Partisan Incentives to Politicize the
Judiciary (Harvard Kennedy Sch. Faculty Research Working Paper SerieRsWP15-001, 2015),
PersonId=280 [https://perma.cc/GNN8-N3SM] (conducting an empirical study of thepoliticization
of the court system).
students came very close to agreeing with the judges in admitting some
other theft crimes, such as shoplifting or grand theft auto. This leads to a ep-r
sumption that the probative value for these crimes is generally higher than
their unfair prejudice.
A more sophisticated critique of the survey is that it is
decontextualized; that is, the judges are assigning levels of probative value and unfair
prejudice in the abstract, without an actual defendant with actual charges in
front of them. Judges may act somewhat differently in actual trial situations,
for a number of reasons. First, they may be more thoughtful and engage in
more critical thinking in making a re-lailfe decision than when they are
simply filling outa survey. Second, the survey results are anonymous,
whereas the judges’ actual rulings are public, and therefore may be uin-fl
enced by how the judge believes she will be perceived by others (current
litigants, future litigants, other judges, etc.). Third, the survey specified that
the prior conviction was “completely dissimilar” to the charge the
hypothetical defendant is currently facing, whereas in an actual trial situation the
prior conviction may be similar to the current charge to varying degrees (for
example, if the defendant is now on trial for firearm possession, a prior
conviction for selling drugs could be seen as related because some jurors
may associate selling drugs with possessing firearms), which should change
how the judge would gauge the level of unfair prejudice. And finally, in the
actual courtroom, judges may be influenced by a wide variety of implicit
biases—based on the race of the defendant, the judge’s own knowledge of
the defendant’s prior record or other aspects of the defendant’s background,
the type of crime the defendant is accused of, the strength of the evidence
against the defendant, or any other factors.151
All of these critiques are legitimate reasons why the survey results may
be inconsistent with how judges actually rule in cases, but most of them are
not reasons to believethat the survey results are inaccurate in their ie-st
mates of probative value and unfair prejudice. Indeed, the second and fourth
of these critiques (the fact that the survey is anonymous and responses are
more likely to be free of implicit biases) are reasons to believe that the
sruvey results more accurately reflect the actual probative value and unfair
prejudice than the judges’ actual rulings. The first critique—that judges may
be more thoughtful and analyitcal in actual cases—cuts the other way,
although it is possible that judges will in fact be more thoughtful when they
are forced to explicitly rate the probative value and unfair prejudice of a
prior conviction, as they are in the survey. The third critique does not really
151 Thanks to Professor Anna Roberts for this critiqueS.ee Roberts, Implicit Stereotyping,
supra note 1, at 861–68.
affect the accuracy of the survey results; it merely points out that real trials
are more complex than the survey can model.
Having acknowledged the differences between how judges maey- r
spond in the survey and how they may rule in actual cases, we can now turn
to an analysis of actual cases. In the next section, we will examine how
judges rule on Rule 609 cases in the courtroom, a nthden compare those
rulings to our survey results.
III. RULE 609 IN THE COURTROOM
If the survey’s averages were a completely accurate indicator of how
judges applied the Rule 609(a)(1)(B) balancing test, then judges would
routinely admit all theft crimes and exclude all other crimes (assuming, as the
survey did, that the prior conviction and the crime for whichthe defendant
is on trial are completely dissimilar). For many critics of Rule 609, even
this state of affairs would be unacceptable: to them, even the admission of a
prior theft conviction against a defendant accused of a completely unrelated
crime is an injustice.152 For these critics, the marginal probative value of
prior convictions of theft to prove lack of credibility is too low, and the
unfair prejudice of the jury hearing about the prior crime (even with a limiting
instruction) is too high.153
This is, in the end, a judgment call: there is no empirical evidence of
how indicative a prior felony theft conviction is to prove lack of credibility
on the stand, and there is no real way of knowing the extent to which jurors
will misuse the conviction as evidence of criminal propensity. It is worth
noting, however, that surveys of the students (who had just learned about
the rule a day or two before taking the survey) and of judges (who had years
of experience applying the rule in their courtrooms) indicate emexetlry
strong support for the proposition that, for theft crimes at least, the
probative value of the prior conviction does outweigh its unfair prejudice.
Of course, that result only takes into consideration theaverages of all
the responses; some judges idnicated they would admit almost any prior
conviction, whereas others indicated that they would admit almost no prior
conviction.154 This is unavoidable for any rule that gives such a broad level
152 See Friedman, supra note 1, at 638 (arguing for the abolition of character impeachment of
a criminal defendant); Roberts, Conviction, supra note 1, at 2018–36 (discussing efforts ot
abolition or reform Rule 609).
153 See Hornstein, supra note 1, at 15–19 (discussing marginal probative value compared to
the high prejudicial effects of prior impeachment evidence); Roberts, Conviction, supra note 1, at
1997 (discussing marginal probative value).
154 This diversity is reflected in the standard deviations for each response, as reported in
pApendix A, infra.
of discretion to the decision maker, but it could be used as anargument for
reforming the rule. Even if the results on average seem sensible (admitting
all theft crimes and precluding all other crimes), if a substantial minority of
judges admit convictions that cause a very high level of unfair prejudice,
the rule may do more harm than good.
Furthermore, the survey itself asks judges how they would respond in
a hypothetical case. It could be true that in practice, when deciding actual
cases, judges behave differently. There are certainly isolated examples of
trial judges making questionable decisions regarding Rule 609(a)(1)(B1)5.5
But to support an argument that this decision should be taken out of the
hands of judges altogether, critics of the rule would need to have evidence
that judges misapply the balancing test often enough that they routinely
admit convictions whose unfair prejudice outweighs their probative value.
So the last piece of the empirical data to consider is: how are trial judges
actually applying this rule in practice?
For the most part it seems that fdeeral trial judges do not make
qusetionable Rule 609(a)(1)(B) decisions. An analysis of over one hundred and
twenty federal district court cases156 indicates that this type of abuse is not
widespread. As would be expected from our survey results, trial
courtjudges frequently admit convictions of theft crimes such as possession of stolen
property, larceny, and robbery. On the end of the spectrum, trial courts are
much less likely to admit prior convictions for drug sale, firearm offenses,
and assault, presumably because they are much less probative of credibility.
Below are the results for the ten most common crime15s7. There are three
columns for admissibility: the first column indicates whether the judge
ad155 For example, in a recent case in Tennessee in whicthe defendant was charged with
possessing and selling cocaine, the trial judge, with very little analysis, ruled in favor of admiatting
defendant’s prior convictions for drug possession, drug sale, and aggravated assault to impeach the
defendant if he testified. United States v. Sneed, No. 3:14 CR 00159, 2016 WL 4191683, at *2 (M.D.
Tenn. Aug. 9, 2016) (“If Sneed chooses to testimony [sic], his credibility will be a central issue. We
will allow impeachment using the three underlying felonies due to their anticipated probativel- va
156 To conduct this analysis, we reviewed every federal district court case on Westlaw over
the past twenty-two years that involved a Rule 609(a)(1)(B) motion, including the unpublished
cases. This, of course, does not represent every possible Rule 609(a)(1)(B) decision, because
many judges will make a Rule 609(a)(1)(B) ruling without reporting the decision to Westlaw. It is
possible that this limitation of our empirical analysis may skew the result—sunreported decisions
may be less thoughtful or contain less legal analysis than reported decisions, leading to a different
pattern of admissibility. A more comprehensive survey of every Rule 609(a)(1)(B) decisionn,- i
cluding both reported and unreported decisions, however, would be extremely resource-intensive
and may in fact not be possible.
157 The overall sample included thirty-one different crimes, but most of them only arose once
or twice, which was not enough times to indicate an accurate pattern of admission. These ten
crimes were the only crimes that appeared four or more times in our sample.
mitted the fact of the felony conviction and allowed the prosecutor to reveal
the name of the crime; the second column indicates how often the judge
admitted the fact of the conviction but did not allow the prosecutor to reveal
the name of the crime; and the third column indicates the total percentage of
times that the judge admitted the prior conviction.
A number of aspects of these results stand out. First, contrary to the
argument by some Rule 609 critics that judges routinely admit prior concv-i
tions to impeach criminal defendants, the data indicate that judges only
ad158 This Table is permanently available ahtttp://www.bc.edu/content/dam/bc1/schools/law/
mitted two-thirds of these prior convictions, and the jury is told the name of
the prior conviction only about half the time. This indicates that judges are
not automatically admitting prior convictions, but that they are taking the
balancing test seriously. Furthermore, the results from the courtroom are
broadly consistent with the results of hte student survey and the judge
survey: judges admitted theft crimes by name far more often than other crimes
(other than crimes of drug possession, which will be discussed below)
whereas crimes of violence are admitted far less often.159
Our analysis of thecases indicate that most judges were acutely aware
of the danger that exists when a prior conviction was similar to the crime for
which the defendant was currently on tria1l6.0 For example, of the nineteen
drug sale convictions that were precluded, twelve of them were precluded at
least in part because of the similarity between the prior conviction and the
current charges. The same reasoning supported precluding five of the eight
robbery charges that were precluded. In four other cases, the trial judge ep-r
cluded a prior conviction involving a drug crime because the defendant was
currently on trial for a firearms charge and juries might associate drug crimes
with firearms.161 In another case, the trial judge precluded a prior conviction
for grand larceny because the defendant was currently on trial for a firearms
offense and the judge was concerned that juries may not know the elements
of grand larceny and incorrectly speculate that it involves the use of ae-fir
arm.162 Finally, with regard to prior drug sales, judges frequently compromise
by admitting the fact and the date of a prior conviction without allowing the
prosecutor to elicit the specific name of the crime; often the judge explained
that this was because of the similarity between the prior conviction anhde t
crime for which the defendant was now being charged.
159 See supra notes 141–146 and accompanying text (discussing the results of the survey).
160 Most circuit courts require trial judges to apply a five-factor balancing test when
determining admissibility under Rule 609(a)(2). Although the factors differ slightly from circuit to circuit,
a common factor is the similarity between the charged crime and the prior conviction—the omre
similar the prior conviction, the more likely it will be that the jury will use the prior conviction
improperly as evidence of propensity to commit the crime in questionS.ee, e.g., United States v.
Hernandez, 106 F.3d 737, 739–40 (7th Cir. 1997).
161 See United States v. Wilson, No. 15-c9r-4, 2016 WL 2996900, at *2(D.N.J. May 23,
2016) (“While the offenses are not identical, they are similar because, in this matter, an individual
juror may associate Defendant’s current charge of illegally possessing a firearm as a convicted
felon with his prior conviction for distributing heroin.”); United States v. Figueroa, No. 15-0098,
2016 WL 126369, at *2 (D.N.J. Jan. 11, 2016)(reasoning that an individual juror may associate
drugs and guns); United States v. Vasquez, 840 F. Supp. 2d 564, 569 (E.D.N.Y. 2011) (concluding
that drug possession is not sufficiently “veracityr-elated” to be more probative than prejudicial;)
United States v. White, No. 08-cr-0682, 2009 WL 4730234, at *4 (E.D.N.Y. Dec. 4, 2009).
162 United States v. Alexander, No. CR-04-64-B-W, 2005 WL 2175169, at *2 (D. Me. Sept. 6,
2005). The trial occurred in Maine, and Maine did not have a crime for grand larceny, so the trial
judge was concerned that the jurors would be too unfamiliar with the crime.
And yet the statistics from the courtroom do show some troubling
trends. Judges admit crimes of violence at an oddly high rate: over half of
the prior convictions for assau-ltype crimes were admitte,d even though
our survey results show that they have close to the lowest level of probative
value for credibility and a relatively high level of unfair prejudice.163 Judges
also admitted three quarters of the prior convictions for drug possession,
even though surveys indicated that the unfair prejudice of that crime is far
higher than the probative value for credibility. Perhaps most troubling of all
is that in approximately 18% of the cases, the prior conviction was admt-i
ted—including the name of the crim—eeven though it was identical or
nearly identical to the crime for which the defendant was currently oni-tr
al.164 This implies that a substantial minority of the judges admit prior
cnovictions in which the unfair prejudice almost certainly outweighs the
probative value. This is not surprising: the survey indicated a number of outlying
judges who would admit nearly every prior conviction; thus, we can assume
that some judges will be very liberal in their courtroom rulings. Neveret-h
less, evidence that such a significant percentage of rulings are so out of step
with the mainstream consensus provides support for the argument that the
rule may need to be amended to avoid these types of rulings.
IV. A NEW RULE 609
Any attempt to reform Rule 609 must consider its placein the overall
context of impeachment evidence and propensity evidence. Many critics
advocate barring the use of prior convictions when used to impeach
criimnal defendants—that is, abolishing Rule 609(a)(1)(B) entirely and
amending Rule 609(a)(2) (which automatically admits all crimes of falsity) to
ensure that it does not apply to criminal defendants1.65 But this change on its
own would create inconsistencies in the law: prior convictions would still
be allowed to impeach any other witness, whereas other tpyes of dishonest
conduct could still be used to impeach criminal defendants under Rule 608.
Furthermore, abolishing Rule 609(a)(1)(B) entirely would mean that
prior convictions could never be used to impeach, no matter how numerous
they were and no matter what type of crime the defendant had committed,
primarily under the theory that the unfair prejudice of such evidence almost
163 See supra note 136 and accompanying text (discussing the probative value for credibility
and level of unfair prejudice of assault-type crimes in Tables 2 and 3).
164 In one notable case, the defendant was on trial for possession of drugs with the intent to
sell them, and the judge admitted four prior convictions for drug possession and one prior
convcition for drug possession with intent to sell. United States v. Alexander, No. 11 CR 114,8-2014
WL 64124, at *6–8 (N.D. Ill. Jan. 8, 2014).
165 See Roberts, Conviction, supra note 1, at 2018–36 (discussing efforts to abolish Rule 609).
always outweighs its probative value.166 As we can see from both the survey
data and the analysis of courtroom decisions, this perspective is
dramatically at odds with how judges view this evidence. It is certainly possible, as
some critics contend, that the vast majority of judges (not to mention the
majority of law students) are misguided, and that prior convictions should
never be admitted against criminal defendants under any circumstance1s6.7
But it seems more likely that these convictions are at least occasionally
more useful to a jury than they are harmful, which would lead us to seek out
a more modest reform of Rule 60—9 one which, incidentally, is far more
likely to be politically feasible than outright abolition.168
Rule 609(a)(1)(B) already contains a balancing test, which
iosstensibly very favorable to the defendant: prior convictions are only supposed to
be admitted if their probative value outweighs their unfair prejudic16e9. If
judges properly applied this test in every case, it would be hard to make a
case for changing the rule. The concern is that in practice, judges (or at least
some judges) do not apply the balancing test properly: theyeither
overestimate the probative value of the evidence, underestimate its unfair prejudice,
do both, or perhaps ignore the balancing test altogether. The response to these
problems should not be to abolish the rule entirely, but to provide judges with
more guidance in the text of the rule so that they follow the abnacling test
more accurately and more consistently.
After examining the empirical data, we can reach two tentative
conclusions. First, the surveys of students and judges reach a consensus that for
some prior convictions—those that involve crimes of the—ftthe probative
166 See FED. R. EVID. 609(a)(1)(B).
167 See generally, e.g., Friedman, supra note 1 (arguing for the abolition of Rule 609 on the
grounds that the prejudicial effect of prior conviction evidence always outweighs the probative
value of such evidence); Hornsteins,upra note 1 (same); Spector, Impeaching the Defendan,t
supra note 1 (same).
168 But see Roberts, Conviction, supra note 1, at 2016–36 (tracing the successful attempts to
reform Rule 609 in three different state evidence codes).
169 In contrast, if Rule 609(a)(2) did not exist, the admissibility of a defendant’s prior
convcitions would be assessed under Rule 608, which would admit the prior conviction as long as the
unfair prejudice did not substantially outweigh its probative value. This would result in a much
larger number of prior convictions being admitted. Of course, those who advocate abolishing Rule
609(a)(2) are aware of this fact, and many of them propose replacing the rule with another rule
that bans prior conviction evidence to impeach criminal defendants in any context, thus carving
out an exception to Rule 608.See, e.g., Roberts, Conviction, supra note 1, at 2036 (proposing the
following change to the statutory language: “In a criminal case whethree defendant takes the
stand, the prosecution shall not ask the defendant or introduce evidence as to whether the
defendant has been convicted of a crime for the purpose of attacking the defendant’s credibility. If the
defendant denies the existence of aconviction, that denial may be contradicted by evidence that
the conviction exists.”).
value for impeachment outweighs the unfair prejudice1.70 The surveys also
reach a consensus that some prior convictions—crimes of violence, sex
crimes, and drug crimes—have very little probative value for impeachment
and a very high level of unfair prejudice.171
Second, our analysis of trial court judges applying Rule 609 in the
courtroom shows rulings that are broadly consistent with the survey results:
crimes of theft are very often admitted, whereas crimes of violence are not.
And yet some glaring inconsistencies exist: specifically, crimes of drug
possession are admitted to impeach even though the survey states that they
have almost no impeachment value, and a substantial nmoriity of judges
admit prior convictions even though they are identical to the crime for
which the defendant is now on trial.
These two conclusions point us in a similar direction: we need to
amend Rule 609 to provide more guidance to those trial judges who are
improperly admitting prior convictions that do not in fact pass the balancing
test. We can provide that guidance by explicitly stating that theft crimes can
be admitted for this purpose, but that no other crimes should be considered.
This would be a relatively simple amendment to Rule 609(a),172 which now
Rule 609. Impeachment by Evidence of a Criminal Conviction
(a) In General. The following rules apply to attacking a witness’s
character for truthfulness by evidence of a criminal conviction:
(1) for a crime that, in the convicting jurisdiction, was punishable
by death or by imprisonment for more than one year, the
(A) must be admitted, subject to Rule 403, in a civil case or in a
criminal case in which the witness is not a defendant; and
170 In the survey, crimes of theft included burglary, carjacking, embezzlement, grand theft
auto, shoplifting, and insider trading.
171 Of course, just because students and judges reach a consensus on these issues does not
mean that they are correc.t It could be, as the Rule 609 critics might contend, that the probative
value of prior convictions is never sufficient to overcome the substantial unfair prejudicial effect
that they have on the jury. But there is naopparent way to measure the true probative value or
unfair prejudice of this kind of evidence; thus, the opinions of those who are first exposed to the
rule and of those who apply the rule on a regular basis are at least useful data points in evaluating
172 See FED. R. EVID. 609(a). The rest of Rule 609 deals with convictions that are over ten
years old, convictions that have been pardoned or annulled, juvenile convictions, and convictions
that are being appealed, all of which are issues outside the scope of thisArticle. See id. R. 609(b)–
(B) must be admitted in a criminal case in which the witness is a
defendant, if the probative value of the evidence outweighs its
prejudicial effect to that defendant; and
(2) for any crime regardless of the punishment, the evidence must
be admitted if the court can readily determine that establishing the
elements of the crime required provin—gor the witness’s
admitting—a dishonest act or false statement.
The proposed amendment would be to Rule 609(a)(1)(B), which would now
(B) may be admitted in a criminal case in which the witness is a
defendant, only if the crime involves an element of thefte, - r
ceiving stolen property, or similar criminal activity, andif the
probative value of the evidence outweighs its prejudicial effect to
that defendant; and
This would limit impeachment by prior conviction for criminael- d
fendants to two types of crimes. First, under our amended Rule 609(a)(1),
felonies that contain an element of theft would be admissible against a
criminal defendant if the probative value outweighed the unfair prejudice, and
admissible against any other witness if it passed the Rule 403 test. Second,
under the existing Rule 609(a)(2), felonies that contain an element of falsity
would be automatically admitted in all cases.173
This amendment would serve a number of important purposes. First, it
would bar the admission of prior crimes that have (according to our survey
173 This proposal parallels Michigan’s Rule 609, which automatically admits crimes of falsity,
and then admits theft crimes if they are felonies and (if the witness is a criminal defendatnhte)
probative value of the evidence outweighs its prejudicial effectS.ee MICH. R. EVID. 609(a). It is
worth noting that this amendment would bring the federal Rule 609 closer in content to many
states’ Rules of Evidence. In twenty-two states, theft crimes are considered “crimes of falsity” and
are therefore automatically admissible under the state’s version of Rule 609. See Huffman v. State,
706 So. 2d 808, 813 (Ala. Crim. App. 1997); Richardson v. State, 579 P.2d 1372, 1376–77
(Alaska 1978); Webster v. State, 680 S.W.2d 906, 908(Ark. 1984); Webb v. State, 663 A.2d 452, 461
(Del. 1995); State v. Page, 449 So. 2d 813, 815–16 (Fla. 1984); People v. Spates, 395 N.E.2d 563,
568–69 (Ill. 1979); Newman v. State, 719 N.E.2d 832, 836 (Ind. Ct. App. 1999); State v.
Harrington, 800 N.W.2d 46, 51 (Iowa 2011); State v. Mahkuk, 551 P.2d869, 872–73 (Kan. 1976); State
v. Grover, 518 A.2d 1039, 1041 (Me. 1986); State v. Brown, 621 N.E.2d 447, 454 (Ohio Ct. App.
1993); Cline v. State, 782 P.2d 399, 400 (Okla. Crim. App. 1989); State v. Gallant, 764 P.2d 920,
922–23 (Or. 1988); Commonwealth v. Kyle, 533 A.2d 120, 123 (Pa. Super. Ct. 1987); State v.
Shaw, 492 S.E.2d 402, 403–04 (S.C. Ct. App. 1997); State v. Butler, 626 S.W.2d 6, 11 (Tenn.
1981); Bello v. State, No. 05–14–00284–CR, 2015 WL 2358173, at *2–3 (Tex. Ct. App. 2015);
State v. Ray, 806 P.2d 1220, 1234(Wash. 1991) (en banc); State v. McDaniel, 560 S.E.2d 484,
490 (W. Va. 2001).The proposed amendment would not automatically admit theft crimes as
crimes of falsity, but it recognizes (as nearly half the state courts do) that these crimes are
particularly probative for demonstrating propensity to lie on the stand.
results) very low probative value and very high unfapirejudice, such as
crimes of violence, drug possession, or drug sale. As we saw in our analysis
of actual cases, there are many judges who admit these prior convictions at
significant rates (though generally not as high a rate as the theft crimes),
even though the survey respondents indicated they should almost never be
This amendment ensures that Rule 609 stays consistent with
608(b), which allows parties to impeach with prior dishonest action1s7.4
According to the survey results, theft crimes have a high probative value for
dishonesty; thus, many of them would be admissible under Rule 608(b) even in
the absence of Rule 609. On the other hand, the survey indicates that other
crimes, such as assault and drug crimes, have a very low level of bparotive
value for dishonesty and a relatively high level of unfair prejudic;ethus, the
use of these prior convictions to impeach criminal defendants would almost
certainly be barred by Rule 608(b) in the absence of Rule 609.175
Because our survey only compared the probative value and unfair
prejudice for criminal defendants, it provides no support for amending Rule
609(a)(1)(A), which covers impeaching witnesses other than criminale-d
fendants.176 The probative value of prior convictions is the same regardless
of the identity of the witness being impeached, but the unfair prejudice to
the party who called the witness will be much lower if the witness is not the
criminal defendant. Thus, it could be that even a crime of violence or a drug
crime, which have a very low probative value, may have such a low level of
unfair prejudice that the unfair prejudice does not substantially outweigh
that low probative value.
Our analysis of the case law shows thtahtis would be a significant
change in the use of criminal convictions to impeach criminal defendants.
Judges currently admit about 68% of prior convictions for crimes unrelated
to theft177—even though the survey results indicate a strong consensus
among judges that the unfair prejudice of these crimes outweighs their
porbative value. Our proposed rule would automatically bar all of these prior
convictions. Furthermore, non-theft crimes are offered to impeach criminal
174 See FED. R. EVID. 608(b).
175 One other possible amendment would be to codify the common law rule that similarity
between the prior conviction and the current charge should bea factor weighing against
admissibility, because our analysis of case law shows some judges admitting prior crimes that are ideni-t
cal to the current charges.See United States v. Hernandez, 106 F.3d 737, 739–40 (7th Cir. 1997)
(setting out that five factors that should be considered when deciding whether to admit prior
176 See FED. R. EVID. 609(a)(1)(A).
177 This percentage includes instances in which the judge admits the prior conviction but does
not give the jury the name of the prior crime.
defendants far more often than theft crimes are offered,178 so a large
proportion of prior convictions that are currently admitted to impeach would be
Future reforms to Rule 609 may depend on further empirical analysis.
For example, Rule 609(a)(2) automatically admits any crime of falsity180—
even if the prior conviction is not a felony, even it if occurred eight or nine
years before, and even if it is identical to the crime currently being charged.
This probably leads to some crimes of falsity being admitted even though
the unfair prejudice of the prior conviction substantially outweighs its
porbative value. And for witnesses who are not criminal defendants, more work
needs to be done to measure the unfair prejudice to the party who called the
witness that is caused by admitting the witness’s prior convictions. Iin-tu
tively, it would seem that the unfair prejudice is less than if the criminal
defendant himself is being impeached, but the degree to which that is true
could still be measured.
For decades, both critics anddefenders of Rule 609 have been talking
past each other, each making claims about the usefulness of prior conc-vi
tions to the jury, and the unfair prejudice that their admission may cause.
The legislators who originally passed Rule 609 were certain that this i-ev
dence was extremely probative for jurors to know about; the scholars who
have uniformly criticized the rule ever since its passage are equally certain
that the implementation of the rule almost always does more harm than
good. This Article has attempted to bring some empirical evidence to bear
on this decades-old dispute, and it uses that evidence to bring a more
modest, targeted reform that will limit abuse of the rule but still ensure that the
jury learns about the most useful types of criminal convictions.
178 Of the 252 prior convictions that we examined in our analysis, 75% were crimes that did
not involve theft. This would mean that in 75% of the cases in which prosecutors currently request
that prior convictions be admitted the proposed Rule 609 would bar admissibility of those
179 Our data sample excluded cases in which the prosecutor offered a crime of falsity under
Rule 609(a)(2), because those lay outside the scope of our survey. The proposed amendment
would not affect the admissibility of these conviction—sthey would still be automatically
admsisible, as they are now. The proposed amendment would also not affect the admissibility of the
25% of prior convictions from our study that involve a crime of thef—tthey would still be subject
to the same balancing test, and presumably would still be admitted approximately at a 72% rate, as
they are now.
180 For this rule to apply, the convictionmust have occurred within ten years of the trial.See
FED. R. EVID. 609(b).
The Survey Instrument Rule 609 Poll
Assume that a defendant is on trial and will take the stand, and the jury
will hear about one of his prior convictions for impeachment purposes. On a
scale of 0-100, how probative do you think each of the following crimes
would be to prove that the defendant has a tendency to lie and is therefore
more likely to be lying on the stand? And on a scale of 0-100, how unfairly
prejudicial do you find each of the following crimes? Assume that
alclonvictions took place five years ago. Also assume that the crime for which the
defendant is now on trial is completely dissimilar from the prior conviction.
181 These Tables are permanently available ahtttp://www.bc.edu/content/dam/bc1/schools/
APPENDIX B—RAW DATA
Comments by Judges
I am in receipt of your request that I complete a brief survey withe-r
gard to the admissibility (or otherwise) of prior criminal convictions of a
I simply do not believe that I can answer this survey in any way, given
that the decision to admit a prior conviction, assuming the legal
prerequisites are present, depends greatly on thecontext in which the prior
convcition is offered, based upon the testimony of the defendant which the prior
conviction seeks to impeach, as well as upon the overall context of the case,
as it has developed, including the charge on which the defendant is
presently standing trial.
If I have, somehow, missed the purpose of the survey or, for that
mtater, have misconstrued the nature of the question asked, please feel free to
so advise, and I will do the very best I can ainnswering your questions,
based upon your explanation.
* * * *
I am unable to participate in your Rule 609 Judicial Survey, but I don’t
want to ignore your efforts. I think you should know my reasons for not
I think the reduction of judicial decision making to quantitative adat
based on hypothetical facts is inappropriate. I never decide questiones- r
garding the admissibility of prior convictions based on Rule 609 without a
full consideration of the context in which the question arises. A mechanical
application is something I never do. There may be a case in which the prior
conviction is for assault causing minor injury in which I would admit the
fact of the conviction and in yet another case charging the same offense I
would not. The same applies to the other offenses you list. Much of the
xeercise of my discretion depends on the degree of distortion from the truth
that either or both sides is presenting in the case itself. If the prosecution’s
case is weak, I probably wouldn’t admit it; if the defense is fanciful or
generally misleading, I probably would. If both were occurring, I would weigh
the question very carefully knowing full well it is a close call.
As an additional point, I never use -scoalled standard instructions. I
use pattern jury instructions as one of many sourcfeosr constructing my
own instructions tailored to each case, and I use these instructions
repeatedly throughout the trial, fromvoir dire to the conclusion of
I know that you are trying to establish patterns, but I don’t think or
decide in patterns. As an example, in over 39 years on the bench, I have never
sentenced according to the Sentencing Guidelines and to apply them in rote
fashion would in my view constitute a violation of my oath of office.
Sometimes the sentences I impose serendipitously coincide with a Guideline
calculation, but I have never imposed a sentence because it was formulated or
dictated by the Guidelines.
I sincerely hope this letter is of use to you.
* * * *
I cannot assign percentages. This demonstrates what I might let
someone use for impeachment, but it depends on the case. [This judge used
checkmarks instead of numbers – checking in the probative value column if
s/he might admit the conviction, and the unfair prejudice if s/he would
This is impossible to answer without more information regarding the
crime for which the defendant is on trial. There is no scenario, in my view,
where the prior conviction is probative. Jury prior conviction is prejudicial.
“Felony only” means that the opposing party would only be able to ask
if the declarant was convicted of aFelony without referring to the specific
charge. [This respondent wrote “Felony only” on the survey sheet any time
s/he gave a prior conviction a probative value greater than zero].
Please note several of these examples do not appear to be felonies or
crimes involving dishonesty or false statement, e.g. underage drinking,
prostitution, marijuana possession, etc. Thus, a 609 analysis is impossible.
Best wishes as you continue to strive to assist our profession in its
search for justice. Under the facts underlined above, I would not permit any
of the suggested criminal convictions listed to be used to impeach the
defendant’s testimony. The relative harm caused by the crimes is not relevant
to my determination.
I was a California State Court judge before becoming a federal judge.
There is significant amounts of case law on this topic in California which I
continue to use as a benchmark. If previously deemed admissible and within
10 years, I let it in. If not, I generally do not but will take argument.
More facts are really needed for better analysis—answer is like answer
to most evidentiary questions—it depends.
If the limiting instruction is given, I believe the fact that the defendant
has a prior felony conviction is always probative on the issue of his crie-d
bility, and does not unfairly prejudice the defendant’s right to a fair trial.
Here, we are not told what the prior conviction was, so I cannot say whether
I would give the “sanitizing” instruction, butgenerally, that would depend
upon how serious the prior felony was and how similar it might be to the
crime or crimes charged.
My basis for determining probative value is based on experience in
observing persons with like crimes who more likely than not will be dishonest
in their dealings with the court and the judicial system at one time or
anohter. Having said that it is one factor to take into account and be assessed
along with other person specific factors that are typically taken into account
when assessing credibility. In that, the final determination may be that the
person with the most heinous of past crimes at the time being assessed may
be absolutely truthful, while the person with the least serious conviction in
the past on the occasion being tested on a credibility issue may be
determined to be completely dishonest.
Sanitizing the specifics often helps tip the balance.
It seems pretty artificial to opine not knowing the nature of the instant
offence. Also, the sliders [the tool used on the onlinesurvey] are tedious to
work. I don’t have time to go back and make little adjustments. They are not
more than rough approximations.
53 Roberts, Conviction, supra note 1 , at 1992 (quoting Dannye W. Holley, Federalism Gone Far Astray from Policy and Constitutional Concerns: The Admission of Convictions to Impeach by State's Rules- 1990 -2004 , 2 TENN. J.L. & POL 'Y 239 , 304 - 05 ( 2005 )).
54 Roberts, Conviction, supra note 1 , at 1992- 93 .