¶1 Petitioner Bryan Allen challenges his felony harassment conviction, raising three issues. The primary issue involves whether the trial court erred by not instructing the jury on the potential fallibility of cross-racial eyewitness identification. Based on the facts of this case, Allen cannot show the trial court violated his constitutional rights by refusing to give the cautionary instruction. A second issue involves whether the “true threat” requirement is an essential element of a harassment statute that must be pleaded in the information and included in the to-convict instruction. A third issue involves prosecutorial misconduct. The Court of Appeals rejected the arguments raised. We affirm the Court of Appeals.

Facts

¶2 Gerald Kovacs, who is white, was walking near the University of Washington at dusk when he was approached *614by two young African American men who offered to sell Kovacs marijuana. Irritated, he told them to “F[uck] off.” Verbatim Report of Proceedings (VRP) (Oct. 21, 2009) at 8. The men screamed and cursed Kovacs, and then followed him. One of the men told Kovacs, “I’m going to kill you, you B[itch],” and lifted up his shirt to display what Kovacs believed to be a gun. VRP (Oct. 21, 2009) at 11. Kovacs ran to the nearest gas station and called the police.

¶3 During the 911 call, Kovacs described the man with the gun as an African American in his mid-20s, wearing a black hooded sweatshirt, a hat, and big, gold-framed sunglasses. Kovacs also described the man as being around 5'9" and between 210-220 pounds. He described the other man as an African American in his teens, around 5'5", wearing a “red kind of shirt,” though he could not remember the color exactly. VRP (Oct. 22, 2009) at 4. Several minutes later, based on Kovacs’ description, a University of Washington patrol officer attempted to stop two African American men near the scene of the crime. One of the men, wearing a white T-shirt, fled. The other, Bryan Allen, did not. Seattle City Police detained Allen and Kovacs was transported to the location of the arrest for a showup identification procedure. Though Allen matched Kovacs’ description of the man with the gun as to race, clothing, hat, and sunglasses, physically he was larger at 6'1" and 280 pounds. Kovacs identified Allen as the man who threatened him. The police searched Allen incident to arrest but found no gun, marijuana, or cash.

¶4 The State charged Allen with felony harassment. Prior to trial, Allen requested the court to instruct the jury regarding cross-racial identifications.1 The court refused Allen’s request. No expert testimony on the reliability of *615cross-racial eyewitness testimony was given at trial. The only testimony given on the subject was by Officer Bennett, the officer in charge of directing the showup identification, who, on cross-examination, agreed that he was “aware of studies suggesting that cross [-] racial identifications can be more difficult for people.” VRP (Oct. 21,2009) at 57. He also agreed that “sometimes people of different races will have a more difficult time identifying somebody of a different race,” though he did not see any indication of difficulties in Kovacs’ identification. VRP (Oct. 21, 2009) at 57. Allen’s defense counsel, in closing argument, challenged the reliability of such evidence.

¶5 Later, in rebuttal closing argument, the prosecutor made several comments regarding Kovacs’ character:

So, what’s most important here is whether or not you accept Mr. Kovacs. I would point out to you from the evidence Mr. Kovacs is not a flake. He’s not some derelict. The evidence would show he is a teacher, very passionate about his work. Not only is he a teacher he is a special education] teacher.

VRP (Oct. 21, 2009) at 105-06. Allen objected to this argument on the basis that the State was vouching for Kovacs’ credibility, but the court overruled the objection. The jury found Allen guilty.

¶6 On appeal, Allen argued the trial court’s failure to instruct the jury on the fallibility of cross-racial identifica*616tions violated his rights to present a defense and to due process. Allen also argued for the first time that the true threat requirement of felony harassment is an element that must be pleaded in the information and included in the to-convict instruction, the omission of which in this case resulted in prejudicial error. Finally, he argued prosecutorial misconduct in closing argument denied him a fair trial. The Court of Appeals affirmed the trial court.

Issues

¶7 1. Whether the trial court erred in failing to instruct the jury on the fallibility of cross-racial eyewitness identifications.

¶8 2. Whether the “true threat” requirement of an antiharassment statute is an essential element of the offense that must be pleaded in the information and included in the to-convict instruction.

¶9 3. Whether the prosecutor’s comments on Kovacs character were impermissible and denied Allen a fair trial.

Analysis

1. Cross-Racial Identification Instruction

¶10 Concerns and discussions over the reliability of eyewitness identifications, and more specifically cross-racial eyewitness identifications, have arisen in cases for some time. The United States Supreme Court focused on eyewitness identification problems in United States v. Wade, 388 U.S. 218, 228, 87 S. Ct. 1926, 18 L. Ed. 2d 1149 (1967), noting that the “vagaries of eyewitness identification are well-known; the annals of criminal law are rife with instances of mistaken identification.” The United States District Court for the District of Columbia, in United States v. Telfaire, 152 U.S. App. D.C. 146, 469 F.2d 552 (1972), cited to Wade and discussed the importance of, and need for, a special instruction on the issue of identification in order to *617safeguard the presumption of innocence. The court in Telfaire crafted a special identification instruction for use in future cases to specifically instruct the jury to assess the value of eyewitness testimony based on several considerations.2 This model instruction did not specifically address cross-racial eyewitness identification; however, in his concurring opinion, Chief Judge Bazelon urged that juries be charged specifically on the pitfalls of cross-racial identification and also proposed sample instruction language. Telfaire, 469 F.2d at 559-61 (Bazelon, C.J., concurring).

¶11 After Telfaire, jurisdictions have developed three general approaches to address the problems perceived to be inherent in eyewitness identification testimony. Some have accepted the rationale underlying Telfaire and have required or encouraged a particularized instruction to be given. See People v. Wright, 45 Cal. 3d 1126, 755 P.2d 1049, 248 Cal. Rptr. 600 (1988) (approving a condensed Telfairetype instruction and requiring that such an instruction be given when requested in a case in which identification is a central issue and there is little corroborative evidence); State v. Warren, 230 Kan. 385, 635 P.2d 1236 (1981) (holding that where eyewitness identification is a critical part of the prosecution’s case and there is serious doubt about the reliability of the identification, a cautionary instruction should be given); State v. Henderson, 208 N.J. 208, 27 A.3d 872 (2011) (requiring that an instruction on cross-racial identification be given whenever cross-racial identification is an issue at trial).3

*618¶12 In other jurisdictions, the decision has been left up to the discretion of the trial court. See United States v. Sambrano, 505 F.2d 284, 287 (9th Cir. 1974) (applying an abuse of discretion standard to the alleged error of failing to give an eyewitness identification instruction and holding *619the trial court had not abused its discretion where general instructions given by the court adequately directed the jury’s attention to the identification issue); Wallace v. State, 306 Ga. App. 118, 701 S.E.2d 554 (2010) (holding the trial court did not abuse its discretion in refusing the defendant’s requested jury instruction on the reliability of cross-racial eyewitness identification where, by general instructions, the jury was informed that it was required to determine whether the eyewitness identification was sufficiently reliable to help satisfy the State’s burden of proof and other corroborating evidence existed).

¶13 The final approach adopted by some jurisdictions has been to reject outright a requirement for Telfaire-like instructions. The courts in these jurisdictions have held that the other general instructions on witness credibility and the government’s burden of proof are adequate and/or that the identification instructions impermissibly comment on the evidence. See State v. Valencia, 118 Ariz. 136, 575 P.2d 335 (Ct. App. 1977) (finding that the instruction on credibility of witnesses was sufficient and that part of Telfaire-type instruction constituted a comment on the evidence); Nevius v. State, 101 Nev. 238, 699 P.2d 1053 (1985) (holding that specific eyewitness identification instruction is duplicitous of general instructions on credibility of witnesses and proof beyond a reasonable doubt); State v. Classen, 31 Or. App. 683, 571 P.2d 527 (1977) (holding that specific Telfaire-type instruction overemphasized the identification issue and amounted to a comment on the evidence), rev’d on other grounds, 285 Or. 221, 590 P.2d 1198 (1979).

¶14 Our cases suggest we have aligned somewhere between the second and third categories mentioned above. In State v. Laureano, 101 Wn.2d 745, 682 P.2d 889 (1984), overruled on other grounds by State v. Brown, 111 Wn.2d 124, 132-33, 761 P.2d 588 (1988), adhered to on recons., 113 Wn.2d 520, 529, 782 P.2d 1013, 787 P.2d 906 (1989), we discussed, albeit briefly, a challenge to the trial court’s *620failure to instruct the jury on eyewitness identification, including cross-racial or ethnic eyewitness identification. In that case, we favorably cited to two Court of Appeals cases, State v. Jordan, 17 Wn. App. 542, 564 P.2d 340 (1977), and State v. Edwards, 23 Wn. App. 893, 600 P.2d 566 (1979), and found no reversible error.

¶15 In Jordan, the Court of Appeals reviewed a Telfairetype instruction and held the trial judge did not err in rejecting the instruction. In that case the court recognized “the focus and ‘emphasis’ of the instruction is upon the credibility of identification witnesses.. . . Witness credibility is more properly tested ‘by examination and cross-examination in the forum of the trial court.’ ” Jordan, 17 Wn. App. at 545 (quoting State v. Johnson, 12 Wn. App. 40, 45, 527 P.2d 1324 (1974)). Similarly, in Edwards, the Court of Appeals held the trial judge did not err in refusing an instruction charging the jury that it must “ ‘be satisfied beyond a reasonable doubt of the accuracy of the identification of defendant as the person who committed the offense before you may convict him.’ ” Edwards, 23 Wn. App. at 896. Although the instruction was not so “impermissibly slanted” as the Telfaire-type instruction rejected in Jordan, the court held “it nonetheless calls into question the credibility of particular witnesses.” Edwards, 23 Wn. App. at 896.

¶16 After Laureano, the Court of Appeals continued to reject the requirement for a Telfaire-type instruction on eyewitness identification. In State v. Hall, 40 Wn. App. 162, 697 P.2d 597 (1985), the court noted that although this court has not ruled on whether a cautionary instruction on eyewitness identification testimony is always inappropriate, such instructions have been considered comments on the credibility of the identification witness and, in any case, the “court’s general instructions on the ‘beyond a reasonable doubt’ standard enabled the defendant to argue his theory of the case and to attack the credibility of eyewitnesses.” Hall, 40 Wn. App. at 167; see also State v. Watkins, *62153 Wn. App. 264, 275, 766 P.2d 484 (1989). Thus, both prior to and following Laureano, our cases have held that an instruction on eyewitness identification is not constitutionally required.

¶17 Allen argues our case law, i.e., Laureano and the cases that preceded and followed it, is outdated. He argues the scientific data regarding the unreliability of eyewitness identification, and of cross-racial eyewitness identification in particular, is now irrefutable. He submits that since Telfaire, research and studies exposing problems inherent in eyewitness identification testimony have gained wide acceptance.4 While the State notes that some researchers have questioned the methodology used in the empirical studies in this field, and notes that publication in respected, peer-reviewed journals is not a guarantee of the validity of the underlying work, the State does not provide contrary evidence or research nor seriously question the scientific data relied upon by Allen. Based on this data, Allen asks us to adopt a rule of general application, founded in notions of due process, that in cases involving cross-racial eyewitness identification it is reversible error to fail to instruct on cross-racial identification when requested. Allen argues the world has changed, and we must change along with it. We are not convinced, however, that the constitutionality of our case law on this issue has changed.

*622¶18 A problem with the studies Allen relies upon is that none of them support the conclusion that the giving of a cautionary cross-racial identification instruction solves the purported unreliability of cross-racial eyewitness identification any more than would cross-examination, expert evidence, or arguments to the jury. As the Supreme Court has recognized, the United States Constitution “protects a defendant against a conviction based on evidence of questionable reliability ... by affording the defendant means to persuade the jury that the evidence should be discounted as unworthy of credit.” Perry v. New Hampshire, U.S._, 132 S. Ct. 716, 723, 181 L. Ed. 2d 694 (2012). In Perry, the Supreme Court addressed the issue of whether due process requires judicial inquiry into the reliability of a suggestive eyewitness identification that was not the result of police arrangement and held it does not. As part of its analysis, the Court listed safeguards, built into our adversary system, that caution juries against placing undue weight on eyewitness testimony of questionable reliability, including the right to confront witnesses, the right to counsel, eyewitness jury instructions adopted by many federal and state courts, expert evidence, the government’s burden to prove guilt beyond a reasonable doubt, and state rules of evidence. Many of these safeguards were at work in Allen’s trial.

¶19 For example, a defendant has a right to effective assistance of counsel, who can expose the unreliability in eyewitness’ testimony during cross-examination and focus the jury’s attention on the fallibility of eyewitness identification during opening and closing arguments. Allen’s counsel did just that. On cross-examination he questioned Kovacs regarding his mental state during the encounter, regarding the time of day the encounter took place (dusk), and regarding the discrepancy between Allen’s actual height and weight, and the description Kovacs gave the 911 dispatcher. Allen’s counsel also questioned Officer Bennett regarding the potential suggestiveness of showup identifi*623cations, the problems associated with cross-racial identifications, and the lack of other witnesses to the crime. VRP (Oct. 21, 2009) at 54-59. Then, in closing argument, Allen’s counsel discussed how emotion and stress can affect the reliability of identifications and discussed the risk of police influence on identifications. He further discussed the “dangers of cross-racial identification” and explained how cross-racial identification may have impacted this case. VRP (Oct. 21, 2009) at 94-98.

¶20 The requirement that the State prove a defendant’s guilt beyond a reasonable doubt also protects against convictions based on dubious identification evidence. The jury in Allen’s case was instructed on the State’s burden of proof and on witness credibility generally.5 Taken together, these instructions charged the jury with deciding whether the State has proved beyond a reasonable doubt that Kovacs correctly identified Allen as the man with the gun. In conjunction with competent defense counsel, the instruc*624tions focused the jury’s attention to the issue of identification and the reliability of Kovacs’ testimony.6

¶21 Allen notes that in Perry, the Supreme Court identified eyewitness instructions, adopted by many federal and state courts, as one safeguard of several that can help focus the jury’s attention on the fallibility of eyewitness testimony. Yet the Supreme Court has never required such an instruction. Nor have any of the jurisdictions we have aligned with on this issue reversed a conviction on due process grounds for failure to so instruct. Adopting a rule of general application, as Allen requests, would take us far afield from our jurisprudence and require us to revisit Laureano and the cases following it. Laureano does not support a rule of general application, but neither does it support a rigid prohibition against the giving of a cautionary cross-racial identification instruction. Indeed, such a prohibition would be inconsistent with the abuse of discretion standard, which we applied in Laureano, and which the Court of Appeals has applied in the cases following Laureano.7

¶22 Applying that standard, we find no abuse of discretion here. Providing a cautionary cross-racial identification instruction would not have added to the safeguards operating in Allen’s case, a case involving an eyewitness identification based on general physique, apparel, and sunglasses, and not on facial features. During the 911 call, Kovacs described the man with the gun by approximate age, height, and weight. Beyond these general characteristics, Kovacs’ *625description was limited to the man’s apparel: “black hoodie,” “jeans,” “baseball cap,” and “big sunglasses” with “gold on the frames.” VKP (Oct. 23, 2009) at 3. At trial, Kovacs testified, regarding the showup identification, that

[the police officer] asked me to identify a gentleman standing on the street... with some officers and some other people, and point him out. . . .

He was wearing the exact same clothes that he had on earlier, he was wearing the baseball hat, the black hood[ie], and he had the glasses. . . . And I said, yeah, definitely, that is one hundred percent him.

VRP (Oct. 21, 2009) at 16. Kovacs did not make an in-court identification of Allen. In fact, Kovacs testified Allen looked different at trial because he was not wearing the same clothes. Thus, Kovacs did not base the identification on facial features, specific physical characteristics, or merely the fact that Allen is African American.

¶23 The central premise of cross-race bias, the impetus for giving a cross-racial identification instruction, is that people are better able to remember faces of their own race than those of a different race.8 Kovacs’ identification of Allen was less than ideal, as it was based largely on Allen’s apparel and sunglasses and not on his facial appearance or other characteristics personal to Allen. However, a specific cross-racial identification instruction would not have been helpful in a case like this where the witness/victim’s identification was based on identifying factors unrelated to cross-race bias. Indeed, Allen’s proposed instructions, which alerted jurors to studies showing “it is more difficult to identify members of a different race than members of one’s own,” without explaining the scientific foundation for cross-race bias, would have been misleading and counterproductive under these circumstances.

*626¶24 We decline to adopt a general rule requiring the giving of a cross-racial instruction in cases where cross-racial identification is at issue, and the trial court did not abuse its discretion by refusing to give a cautionary cross-racial jury instruction under the facts of this case. We affirm the Court of Appeals.

2. True Threat

¶25 “The First Amendment, applicable to the States through the Fourteenth Amendment, provides that ‘Congress shall make no law . . . abridging the freedom of speech.’ ” Virginia v. Black, 538 U.S. 343, 358, 123 S. Ct. 1536, 155 L. Ed. 2d 535 (2003) (alteration in original) (quoting U.S. Const, amend. I). While the scope of the First Amendment is broad, it does not extend to unprotected speech, one category of which is “true threats.” A “true threat” is “ ‘a statement made in a context or under such circumstances wherein a reasonable person would foresee that the statement would be interpreted ... as a serious expression of intention to inflict bodily harm upon or to take the life’ of another person.” State v. Kilburn, 151 Wn.2d 36, 43, 84 P.3d 1215 (2004) (alteration in original) (internal quotation marks omitted) (quoting State v. Williams, 144 Wn.2d 197, 207-08, 26 P.3d 890 (2001)). Accordingly, we interpret statutes criminalizing threatening language as proscribing only unprotected true threats.

¶26 Allen was charged in this case with felony harassment.9 He argues that because only true threats may be *627prosecuted, the true threat requirement is an essential element of a harassment statute. It therefore must be included in the information and to-convict instruction,10 which did not happen in this case. Here, the information stated:

That the defendant BRYAN EDWARD ALLEN in King County, Washington, on or about August 6, 2009, knowingly and without lawful authority, did threaten to cause bodily injury immediately or in the future to Gerald Kovacs, by threatening to kill Gerald Kovacs, and the words or conduct did place said person in reasonable fear that the threat would be carried out.

Clerk’s Papers (CP) at 1. The to-convict instruction required the jury to find the following elements beyond a reasonable doubt to convict Allen of the crime of felony harassment:

(1) That on or about August 6,2009, the defendant knowingly threatened to kill Gerald Kovacs immediately or in the future;

(2) That the words or conduct of the defendant placed Gerald Kovacs in reasonable fear that the threat to kill would be carried out;

(3) That the defendant acted without lawful authority; and

(4) That the threat was made or received in the State of Washington.

CP at 21. While the true threat requirement was not included in the information or to-convict instruction, the jury was given a separate instruction defining “true threat”:

*628Threat means to communicate, directly or indirectly, the intent to cause bodily injury in the future to the person threatened or to any other person;

To be a threat, a statement or act must occur in a context or under such circumstances where a reasonable person, in the position of the speaker, would foresee that the statement or act would be interpreted as a serious expression of intention to carry out the threat rather than as something said in jest or idle talk.

CP at 20.

¶27 We have never held the true threat requirement to be an essential element of a harassment statute. In State v. Johnston, 156 Wn.2d 355, 127 P.3d 707 (2006), we found that Washington’s bomb threat statute, RCW 9.61.160, reached a substantial amount of protected speech. We therefore construed the statute to avoid an overbreadth problem by limiting it to true threats. We held the trial court erred by giving an instruction that defined “true threat” in terms of the reasonable-listener-based standard, rather than the reasonable-speaker-based standard we adopted in Kilburn. Johnston, 156 Wn.2d at 364. By so holding, we implied a properjury instruction defining “true threat” would have been constitutionally sufficient.

128 In State v. Schaler, 169 Wn.2d 274, 236 P.3d 858 (2010), the defendant, charged under the same statute as was Allen, successfully requested a jury instruction requiring the jury to find he subjectively intended to communicate a threat. No true threat instruction was given. Instead, the jury was instructed that “ £ “threat” means to communicate, directly or indirectly, the intent to cause bodily injury immediately or in the future to the person threatened or to any other person.’ ” Schaler, 169 Wn.2d at 285 (quoting CP at 26, 28 (paralleling RCW 9A.04.110(27)(a))). The jury was further instructed that “ £[a] person threatens “knowingly” when the person subjectively intends to communicate a threat.’ ” Schaler, 169 Wn.2d at 285 (quoting CP at 26, 40, 55-56). Thus, under the instructions given, the *629statutory requirement of knowing or even intentional threatening referred only to the conduct and circumstances proscribed, not to the proscribed result (the listener’s fear). We held that in the context of criminalizing speech, the lack of mens rea as to the result is critical and that the reasonable speaker-based standard requires negligence. Because the instructions did not require a mens rea of at least negligence as to result, the jury could have convicted the defendant on something less than a true threat, and the instructions were therefore in error. Schaler, 169 Wn.2d at 286-87. However, we noted that although the instructions erroneously failed to limit the statute’s scope to true threats, the problem was unlikely to arise in future cases, since after Johnston the Washington Pattern Jury Instructions Committee amended the pattern instruction defining “threat” so that it matched the definition of “true threat.” We said that “[clases employing the new instruction defining ‘threat’ will therefore incorporate the constitutional mens rea as to the result.” Schaler, 169 Wn.2d at 288 n.5. That instruction was used in this case.11

¶29 Moreover, the Court of Appeals has repeatedly held the true threat requirement is not an essential element of harassment statutes. In State v. Tellez, 141 Wn. App. 479, 170 P.3d 75 (2007), the defendant, charged with felony telephone harassment based on a threat to kill, claimed the information and to-convict instruction were deficient be*630cause they lacked the requirement of a true threat, an essential element of the crime. The Court of Appeals agreed with the State that “the constitutional concept of ‘true threat’ merely defines and limits the scope of the essential threat element in the felony telephone harassment statute and is not itself an essential element of the crime.” Tellez, 141 Wn. App. at 484. In so holding, the court in Tellez construed our holding in Johnston — that it is error not to give a limiting instruction defining threat to include only true threats — as characterizing the true threat concept as definitional, and not as an essential element of any threatening-language crime. Tellez, 141 Wn. App. at 484. In State v. Atkins, 156 Wn. App. 799, 236 P.3d 897 (2010), the defendant, charged with felony harassment, likewise contended the information and to-convict instruction were deficient for failing to contain the essential element of a true threat. The Court of Appeals found Tellez dispositive, holding that so long as the jury was instructed as to the true threat requirement, the defendant’s First Amendment rights were protected.

¶30 In this case, Allen argued the First Amendment for the first time on appeal. Under the circumstances, no manifest error affecting a constitutional right occurred. The jury was instructed as to the true threat requirement and Allen’s First Amendment rights were protected. Based on Johnston, Schaler, Tellez, and Atkins, we hold failure to include the true threat requirement in the information and to-convict instruction was not error.

3. Prosecutorial Misconduct

¶31 Allen contends the prosecutor committed misconduct by inappropriately vouching for Kovacs’ credibility in closing arguments. In rebuttal closing argument, the prosecutor stated:

So what’s most important here is whether or not you accept Mr. Kovacs. I would point out to you from the evidence Mr. Kovacs is not a flake. He’s not some derelict. The evidence *631would show he is a teacher, very passionate about his work. Not only is he a teacher he is a special education] teacher.

VRP (Oct. 21, 2009) at 105-06. Allen objected but was overruled. The prosecutor then continued:

The evidence will show that he teaches kids that have disabilities. The evidence will show that Kovacs has two master’s degrees. This is a person that was walking on the street minding his own business and within a matter of minutes he’s threatened with death.

VRP (Oct. 21,2009) at 106. It is misconduct for a prosecutor to state a personal belief as to the credibility of a witness. However, prosecutors have wide latitude to argue reasonable inferences from the facts concerning witness credibility, and prejudicial error will not be found unless it is clear and unmistakable that counsel is expressing a personal opinion. State v. Warren, 165 Wn.2d 17, 30, 195 P.3d 940 (2008) (citing State v. Brett, 126 Wn.2d 136, 175, 892 P.2d 29 (1995); State v. Stenson, 132 Wn.2d 668, 727, 940 P.2d 1239 (1997)).

¶32 The Court of Appeals held the trial court did not abuse its discretion in permitting the comments by the prosecutor in this case. We agree. Here, the prosecutor’s argument was based on evidence introduced at trial: Kovacs testified he was a special education teacher; he testified he had a master’s degree in teaching from the University of Washington and a master’s degree in special education teaching from Pacific Lutheran University; and he testified he served four years in the Army National Guard and had two adopted children. VRP (Oct. 21, 2009) at 5-13. The prosecutor did not rely on information not presented to the jury. Instead he drew an inference from the evidence as to why the jury would want to believe Kovacs. The inference was reasonable, based on Kovacs’ testimony, and the statement was not a clear and unmistakable expression of the prosecutor’s personal opinion. The jury was instructed that it was the sole judge of the credibility of each witness, and *632the prosecutor reminded the jury of this fact before making the inference. Under these circumstances, we hold the prosecutor’s comments were not improper.

Conclusion

¶33 Based on the facts of this case, we hold the trial court did not err by failing to provide a cross-racial identification instruction. We further hold that the true threat requirement is not an essential element of felony harassment and that the instruction defining “true threat” in this case safeguarded Allen’s First Amendment rights. Finally, we hold the prosecutor did not commit misconduct in closing argument.

¶34 (concurring) — I agree with the lead opinion that State v. Laureano, 101 Wn.2d 745, 682 P.2d 889 (1984), overruled on other grounds by State v. Brown, 113 Wn.2d 520, 782 P.2d 1013 (1989), and subsequent Court of Appeals’ decisions have rejected the categorical use of a Telfaire12 instruction. I further agree with the lead opinion that the trial court did not abuse its discretion in this case because there was no indication that Gerald Kovacs’ identification of Bryan Allen was based upon facial features or other specific physical characteristics beyond the mere fact that Allen is African American.

¶35 I write separately because I believe in a hypothetical case where a victim makes a cross-racial identification based on a suspect’s facial features, hair, or other physical characteristic implicating race, a trial judge likely would abuse his or her discretion if he or she refused to provide a cross-racial identification instruction. The dissent properly recognizes that cross-examination, expert testimony, and *633closing argument may not provide sufficient safeguards against cross-racial misidentification because the very nature of the problem is that witnesses believe their identification is accurate. Further, as discussed by the dissent, with social science data increasingly casting doubt on the reliability of cross-racial identification, our courts must carefully guard against misidentification.

¶36 However, the dissent’s concerns are misplaced in this case. The identification here simply did not implicate the type of physical characteristics that give rise to erroneous cross-racial identifications or the need for an instruction. Besides reporting the suspect to be African American, Kovacs described the man in race-neutral terms: mid-20s; about 5'9" in height and 210-220 pounds; wearing a black hooded sweatshirt, dark blue jeans, big gold-framed sunglasses, and a dark baseball cap; and possessing a gun. When asked by the 911 operator if the man had any facial hair, Kovacs responded vaguely, “Not that I remember,” signaling a lack of attention to facial features. I Verbatim Report of Proceedings (Oct. 22, 2009) at 5. At trial, when Kovacs described how he identified Allen as the man who had threatened him, he referred to Allen’s hat, clothes, and sunglasses. Indeed, Kovacs did not even mention Allen’s facial features, hair, or other physical characteristics. While one might infer from Kovacs’ estimate of the older suspect’s age and the lengthy encounter between Kovacs and both suspects that Kovacs must have considered facial features at some point, this would be purely speculative. Instead, the record in this case supports the lead opinion’s conclusion that the identification here was based primarily on race-neutral factors.

¶37 Therefore, I agree with the lead opinion that the trial court did not abuse its discretion when it refused to give an instruction on cross-racial identification in this case.

¶38 (concurring in result) — I concur with much in the well reasoned dissent. The cross-racial instruction is correct and will be necessary from time to time to instruct the jury on the dangers of cross-racial identification. However, I join the lead opinion in result because I also agree with Chief Justice Madsen that under the facts of this case we cannot say the trial judge abused her discretion in declining to give the instruction. I also agree with the lead opinion that “true threat” need not be pleaded in the information and included in the “to convict” instruction, though of course the jury must be properly instructed on its meaning.

¶39 I write separately to stress that the lead opinion holds, and I agree, that expert testimony on the weakness of cross-racial identification is admissible when relevant and helpful. Lead opinion at 624 n.6 (citing State v. Cheatam, 150 Wn.2d 626, 646, 81 P.3d 830 (2003)); see also State v. Jaime, 168 Wn.2d 857, 869, 233 P.3d 554 (2010) (Sanders, J., concurring). The recognition that expert testimony is admissible is very important to our justice system, which for so long has relied so heavily upon eyewitness identification to convict and sentence. The American Bar Association reports that “[approximately three-quarters of the more than 200 wrongful convictions in the United States overturned through DNA [deoxyribonucleic acid] testing resulted from eyewitness misidentifications. Of that 77 percent, where race is known, 48 percent of the cases involved cross-racial eyewitness identifications.” Am. BarAss’n, Criminal Justice Section, Report to the House of Delegates 6 (Aug. 2008) (ABA Report)13 (citing Innocence Project Fact Sheets, Eyewitness Misidentification and Facts on Post-Conviction DNA Exonerations). The amici briefs submitted by college *635and university professors, the American Civil Liberties Union of Washington, the Fred T. Korematsu Center for Law and Equality, and the Innocence Network, joined by many others, bring a wealth of research demonstrating the dangers of cross-racial identification, which the State does not deny.

¶40 Unfortunately, the value of any expert testimony will be diluted without an instruction to guide the jury in bringing the expert’s testimony into their deliberations in a reasoned way. We now know that such instructions are necessary to ensure a fair trial. See ABA Report (discussing the need for such an instruction). Indeed, the better practice may be to instruct whenever cross-racial identification is implicated. State v. Henderson4 208 N.J. 208, 27 A.3d 872 (2011) (requiring such instructions). At the very least, I concur with the dissent that trial courts should be required to give the instruction where eyewitness identification is a central issue in a case, there is little evidence corroborating the identification, and the defendant specifically requests the instruction. Dissent at 637. I also stress that we have long rejected the contention that such instructions function as unconstitutional comments on the evidence. State v. Carothers, 84 Wn.2d 256, 267-68, 525 P.2d 731 (1974).

¶41 We must learn from our mistakes; both liberty and justice depends upon it. Given the demonstrated weakness of eye witness testimony in general and cross-racial eyewitness identification in particular, in my view, expert testimony and instruction to the jury on the weakness of cross-racial identifications should be the standard in our courtrooms whenever it would be helpful. I respectfully concur in result.

¶42 (dissenting) — The most important lesson of this case is that every member of this court would *636support giving a cross-racial identification instruction in an appropriate case — but we differ on what constitutes an appropriate case. The four-justice lead opinion declines to adopt a rule of general application defining the circumstances under which an instruction should be given and equally rejects a rule that a cross-racial instruction should never be given. Lead opinion at 624 (“[Our prior decision] in Laureano[14] does not support a rule of general application, but neither does it support a rigid prohibition against the giving of a cautionary cross-racial identification instruction. Indeed, such a prohibition would be inconsistent with the abuse of discretion standard, which we applied in Laureano, and which the Court of Appeals has applied in the cases following Laureano”). The three concurring justices more explicitly state that they would support the instruction in an appropriate case. Concurrence (Chambers, J., joined by Pairhurst, J.) at 634 (“The cross-racial instruction is correct and will be necessary from time to time to instruct the jury on the dangers of cross-racial identification.”); concurrence (Madsen, C.J.) at 632 (“I write separately because I believe in a hypothetical case where a victim makes a cross-racial identification based on a suspect’s facial features, hair, or other physical characteristic implicating race, a trial judge likely would abuse his or her discretion if he or she refused to provide a cross-racial identification instruction.”). The two dissenting justices would offer more specific guidance for determining when an instruction should be given and would reverse this conviction for the failure to give an instruction here.

¶43 Given the unanimous fundamental agreement of this court that cross-racial instructions are appropriate at least some of the time, trial courts and counsel should consider the appropriateness of a cross-racial instruction in any case involving a cross-racial identification.

*637¶44 I now explain the circumstances in which I believe the instruction should be given. I agree with the lead opinion that we need not adopt an across-the-board rule requiring a cross-racial identification instruction in every case potentially raising the issue. But courts should be required to give an instruction where eyewitness identification is a central issue in a case, there is little evidence corroborating the identification, and the defendant specifically asks for an instruction.

I. The lead opinion overlooks a number of critical facts in concluding that an explanatory instruction was not proper

¶45 When all relevant facts are taken into account, it is plain that this is precisely the factual scenario calling for a jury instruction on the dangers of cross-racial identification. Consider the facts:

1. The case involved a cross-racial identification of a black suspect by a white victim, the racial combination accounting for most identification errors. See John P. Rutledge, They All Look Alike: The Inaccuracy of Cross-Racial Identifications, 28 Am. J. Crim. L. 207, 211 (2001) (citingPeople v. McDonald, 37 Cal. 3d 351, 690 P.2d 709, 720, 208 Cal. Rptr. 236 (1984), overruled on other grounds by People v. Mendoza, 23 Cal. 4th 896, 4 P.3d 265, 98 Cal. Rptr. 2d 431 (2000)).

2. There is barely any evidence corroborating the identification of the State’s witness, Gerald Kovacs:

a. The person who was with Bryan Allen when he was stopped did not match the description of the armed suspect’s companion.

b. Allen himself did not match Kovacs’ description, being four or five inches taller and 60 pounds heavier than Kovacs’ description.

c. No gun, weapon, or other item that could be mistaken for a gun was found on Allen.

*638d. No drugs or other evidence were recovered on Allen to corroborate Kovacs’ claim that Allen was selling drugs.

3. The identification occurred under circumstances calling its accuracy into question.

a. The incident occurred at dusk, limiting Kovacs’ ability to identify the suspect.

b. A weapon was involved, making it harder for Kovacs to accurately identify the suspect due to “weapon focus.” Gary L. Wells & Deah S. Quinlivan, Suggestive Eyewitness Identification Procedures and the Supreme Court’s Reliability Test in Light of Eyewitness Science: 30 Years Later, 33 Law & Hum. Behav. 1, 11 (2009).

c. The suspect wore a cap and sunglasses. The use of disguises compromises identification accuracy. Richard A. Wise et al., How to Analyze the Accuracy of Eyewitness Testimony in a Criminal Case, 42 Conn. L. Rev. 435, 456 (2009).

d. The identification was based in part on the cap and sunglasses, which the police required Allen to wear for the showup so that he more closely resembled the suspect. The police also instructed Allen to pull his cap low over his face before the showup.

4. A jury instruction would have been especially helpful in this case given that there was already misinformation about eyewitness testimony in evidence. The State’s chief law enforcement witness claimed that eyewitness evidence is not made more reliable by corroborating evidence and also claimed that the “most reliable” identification is “shortly after a crime has been committed and there is a showup.” Report of Proceedings (Oct. 21, 2009) at 54, 56. Both of these claims are contradicted by prevailing authority. See Nat’l Inst, of Justice, U.S. Dep’t of Justice, Eyewitness Evidence: A Guide for Law Enforcement 27 (1999)

*639(discussing shortcomings of eyewitness identification testimony and the “inherent suggestiveness” of show-up identifications in particular), available at https:// www.ncjrs.gov/pdffiles l/nij/178240.pdf.

The lead opinion is correct that the jury instruction Allen requested would be most helpful where identification was based solely on facial-feature recognition. But the contra-positive does not follow. This case demonstrates that even where identification is not facially oriented, cross-racial identification is fraught with peril and can be highly inaccurate.

II. Cross-racial identification is unreliable

¶46 There is a large body of persuasive scientific research concluding that eyewitness testimony is frequently unreliable. Nearly 80 percent of all wrongful convictions exonerated by DNA (deoxyribonucleic acid) evidence were founded on faulty eyewitness testimony. State v. Riofta, 166 Wn.2d 358, 371, 209 P.3d 467 (2009). Research shows jurors are unable to correctly distinguish between reliable and unreliable eyewitness identification testimony, and jurors consistently over-believe such testimony. Gary L. Wells & Elizabeth A. Olson, Eyewitness Testimony, 54 Ann. Rev. Psychol. 277, 284-85 (2003).

¶47 We also know that the problem is most dramatic when the identifying witness and the suspect are of different races: 40 percent of wrongful convictions exonerated by DNA involve faulty cross-racial identifications, and 36 percent involve whites mistakenly identifying blacks. James M. Doyle, Discounting the Error Costs: Cross-Racial False Alarms in the Culture of Contemporary Criminal Justice, 7 Psychol. Pub. Pol’y & L. 253 (2001). In these cases, a witness’s ability to accurately identify a suspect is severely undermined by the cross-racial effect. Rutledge, supra, at 211. Eyewitnesses are 1.56 times more likely to misidentify a suspect when the identification is cross-racial. See Christian A. Meissner & John C. Brigham, Thirty Years of *640Investigating the Own-Race Bias in Memory for Faces: A Meta-Analytic Review, 7 Psychol. Pub. Pol’y & L. 3, 15 (2001). Compounding the problem, jurors believe cross-racial identification is more accurate than same-race identification when in fact research shows the opposite. Richard S. Schmechel et al., Beyond the Ken? Testing Jurors’ Understanding of Eyewitness Reliability Evidence, 46 Jurimetrics J. 177, 200 (2006).

¶48 The cross-racial identification problem creates a racial disparity at the entry point into the criminal justice system, eventually leading to racial disparity throughout the system. I do not take the lead opinion as disagreeing with the basic premise of this dissent; it is almost beyond dispute that cross-racial identification is problematic. That is not where the lead opinion goes wrong. Instead, the lead opinion makes two faulty assumptions leading it to conclude that no instruction was proper and that we should reject a rule requiring an instruction in certain cases. Both assumptions are unsupported and so is the lead opinion’s position.

III. Existing safeguards against misidentification are inadequate

¶49 The lead opinion first incorrectly assumes that our criminal justice system incorporates adequate safeguards against the dangers of cross-racial identification. This is too optimistic. The lead opinion suggests that cross-examination, expert testimony, and closing argument sufficiently guard against the problem. But cross-examination is a useless tool for educating jurors about cross-racial bias. The very nature of the cross-racial problem is that witnesses are unaware of it; witnesses believe their identification is accurate, making traditional impeachment methods inadequate for ferreting out the truth. Timothy P O’Toole & Giovanna Shay, Manson v. Brathwaite Revisited: Towards a New Rule of Decision for Due Process Challenges to Eyewitness Identification Procedures, 41 Val. U. L. Rev. 109, 135 *641(2006) (concluding that “because the use of suggestive procedures and unreliable identifications almost always occur with eyewitnesses who honestly believe their own mistaken identification, cross-examination is nearly useless”). Social science confirms this logic: studies show that cross-examination fails to increase juror sensitivity to the inaccuracy of eyewitness testimony. See Michael R. Leippe, The Case for Expert Testimony About Eyewitness Memory, 1 Psychol. Pub. Pol’y & L. 909, 923-25 (1995).

¶50 Nor is expert testimony a practical solution. Expert testimony seems like a natural way to educate jurors about cross-racial bias, except that it is far too costly. Experts are both scarce and expensive. See Douglas W. Vick, Poorhouse Justice: Underfunded Indigent Defense Services and Arbitrary Death Sentences, 43 Buff. L. Rev. 329, 375 & n.199 (1995). Most felony defendants in state court are indigent, and public defenders cannot afford to pay expert fees either. See Caroline Wolf Harlow, U.S. Dep’t of Justice, Defense Counsel in Criminal Cases 1 (2000) (“At the end of their case approximately . . . 82% of felony defendants in large State courts were represented by public defenders or assigned counsel.”), available at http://bjs.ojp.usdoj.gov/content/pub/ pdf/dccc.pdf. This being the case, expert testimony for all defendants is not a solution but a pipe dream.

¶51 Likewise, it is a hollow exercise to educate jurors about faulty cross-racial identification in closing argument where an attorney has been deprived of the raw materials integral to building an effective defense. Without evidence or some other form of authority, it is difficult to imagine why jurors would believe defense counsel’s unsupported assertions about cross-racial identification.

¶52 The lead opinion also argues that a reasonable doubt instruction safeguards against the dangers of cross-racial identification, but I do not see how this can be. See lead opinion at 623-24. The implication here seems to be that since the State’s burden of proof is so high in a criminal case, it is not harmful to permit the State to use inaccurate *642and misleading evidence. I reject this notion, failing to see how a reasonable doubt instruction — a feature of every criminal case — protects defendants from the unique inaccuracies of cross-racial identification evidence.

¶53 The lead opinion is too quick to assume that our justice system contains adequate safeguards against inaccurate cross-racial identification.

IV. A jury instruction helps cure the cross-racial problem in certain cases

¶54 The lead opinion is also mistaken in assuming that a jury instruction on the inaccuracy of cross-racial identifications is unhelpful. Jury instructions are in many ways an ideal way to deal with this disparity; the heart of the problem is that jurors believe cross-racial identification is equally or more accurate than same-race identification, when in fact it is far less accurate. Thus, educating jurors is precisely what is called for. Consider the benefits of a jury instruction. First, it costs nothing. Second, jury instructions are focused, concise, and authoritative (jurors hear them from a trial judge, not from a witness called by one side). Third, a jury instruction avoids the problem of dueling experts and eliminates the risk of an expert invading the jury’s role or opining on an eyewitness’s credibility. Fourth, jurors may be more likely to discuss racial differences and the cross-racial problem in deliberation if bolstered by the credibility of an instruction.

¶55 There are benefits beyond the juror box as well. For the courts to recognize that cross-racial eyewitness identification is frequently erroneous would encourage police and prosecutors to approach these identifications cautiously when making charging and investigative decisions. Cf. State v. Martin, 101 Wn.2d 713, 723, 684 P.2d 651 (1984) (concluding same regarding hypnosis evidence). For example, law enforcement personnel might try to find more corroborating evidence where the only link between a suspect and a crime is a cross-racial identification. These *643“upstream effects,” combined with all the other advantages of a jury instruction, demonstrate the unsoundness of the lead opinion’s assumption that a jury instruction is unhelpful.

¶56 Once the lead opinion’s false assumptions are cleared away, little reason remains to reject the palliative measure proposed by the petitioners. I would embrace a version of the rule adopted in other jurisdictions, holding that a court must give the instruction where cross-racial eyewitness identification is a central issue in the case, where there is little corroborating evidence, and where the defendant asks for the instruction. See People v. Wright, 45 Cal. 3d 1126, 755 P.2d 1049, 1059, 248 Cal. Rptr. 600 (1988); see also State v. Long, 721 P2d 483, 492 (Utah 1986). In such cases, there is an impermissible risk of wrongful conviction that is best mitigated by an instruction.

¶57 Because this would be a new rule, I would adopt it on a prospective basis only. See Wood v. Morris, 87 Wn.2d 501, 514, 554 P.2d 1032 (1976). However, in Allen’s case, I would hold that the facts so strongly called for a curative instruction that the trial court committed reversible error by refusing it and that Allen is entitled to a new trial.

V. Conclusion

¶58 The evidence is irrefutable that cross-racial identification is often faulty. Unlike most problems of racial disparity, here we have a simple way to mitigate the damaging effects; a solution that is not only cost-free but also tested, other states having found it workable. We have every reason to adopt this rule and no reason not to. Where we can take simple steps to reduce racial disparity, we should do so rather than turning our backs on the problem.

¶59 I respectfully dissent.