Calvin Minor, appellant, was convicted for armed carjacking, armed robbery, possession of a firearm during a crime of violence (“PFCV”), and unauthorized use of a vehicle (“UUV”)-1 On appeal, Mr. Minor argued that the trial court erred in excluding the testimony of Dr. Ronald Fisher, a professor of psychology, who was proffered to testify about issues related to the reliability of eyewitness identifications. After briefing and argument, we remanded the record and ordered the trial court to conduct a hearing to consider the admission of Dr. Fisher’s proffered testimony. We directed the trial court to consider Benn v. United States, 978 A.2d 1257 (D.C.2009) (Benn II), and Russell v. United States, 17 A.3d 581 (D.C.2011), which were issued subsequent to the trial court’s ruling to exclude the expert testimony but which contain substantial guidance on the factors to be considered in determining the admissibility of expert testimony in this field. Those cases build upon the three-pronged test we established in Dyas v. United States, 376 A.2d 827 (D.C.1977), to determine the admissibility of expert testimony generally:
(1) the subject matter “must be so distinctively related to some science, profession, business or occupation as to be beyond the ken of the average layman” ...; (2) “the witness must have sufficient skill, knowledge, or experience in that field or calling as to make it appear that his opinion or inference will probably aid the trier in his search for truth” ...; and (3) expert testimony is inadmissible “if the state of the pertinent art or scientific knowledge does not permit a reasonable opinion to be asserted even by an expert.”
Dyas, 376 A.2d at 832 (quoting McCormick on Evidence § 13 at 29-31 (E.Cleary, 2d ed.1972)) (emphasis omitted).
On remand, the trial court again ruled that it would exclude Dr. Fisher’s testimony, finding that his testimony would not aid the trier of fact, and therefore would not satisfy the second Dyas factor, because his conclusions were “equivocal” and the underlying studies did not mirror the “real world.” In his challenge to the trial court’s ruling on remand, Mr. Minor argues that the trial court abused its discretion because the underpinnings for its exclusion of the testimony go solely to the weight of the expert testimony and not to its admissibility under Dyas. We agree. We hold that all three prongs of the Dyas test have been satisfied, that the testimony would be more probative than prejudicial, and that its exclusion was not harmless. We therefore reverse and remand for a new trial at which Mr. Minor shall be allowed to present Dr. Fisher’s expert testimony.
*410I. FACTUAL AND PROCEDURAL BACKGROUND
A.The Carjacking
On October 10, 2005, on her way home from a night club, Crystal Nunnley stopped at the 7-Eleven store located at 950 Eastern Avenue in Northeast Washington, D.C. Upon exiting the store, she heard a voice say, “Give me the keys.” When the perpetrator again demanded her keys and also money, Ms. Nunnley looked up to see a “child” pointing a gun at her. She said she had no money, threw her keys on the ground, and ran back inside the store. The perpetrator sped away in Ms. Nunnley’s car, a white Acura, followed by two cars, a burgundy Toyota Célica or Chevy Cavalier and a green car. The incident lasted only “seconds.”
Ms. Nunnley testified that she was “terrified” and “in total shock” and that she was “so frantic” she did not realize she was holding her cell phone, so she used a phone in the 7-Eleven to call 911. When asked by the 911 operator to describe the carjacker, Ms. Nunnley said, “I don’t know,” or “I don’t remember”; she testified at trial that her response merely reflected her exasperation at the operator’s questions. Metropolitan Police Department Detective Laura Aceto and Officers Thomas Caddell and Dale Vernick arrived on the scene a few minutes later. Officer Caddell testified that Ms. Nunnley was “very hysterical.” Ms. Nunnley described the carjacker to the officers as a “young [male], between 16 to 18 years old,” with dark skin, hair done in braids or dreads, and wearing jeans and a jacket with the hood “slightly pulled over his head.”
Approximately one hour later, the police located the burgundy car, and shortly thereafter Ms. Nunnley arrived and identified items scattered on the ground beside the vehicle as hers. At the same time, less than half a block away, police officers stopped a dark-complected black male, with his hair in long twists and wearing a green jacket. Ms. Nunnley did not recognize the man and the officers released him without recording his name. Later that night, Ms. Nunnley’s vehicle was located in an alley less than half a block from where the burgundy car was found.
B.Ms. Nunnley’s Identification of Mr. Minor as the Carjacker
Approximately one week later, Detective James Francis showed Ms. Nunnley three photo arrays. She pointed to Mr. Minor’s picture and said, “[t]hat looks like him.” Detective Francis thought she seemed “very confident” but acknowledged that Ms. Nunnley did not say she was “100 percent sure.” Ms. Nunnley testified that she did not express more confidence in her identification because she was “terrified” of the carjacker’s accomplices who had been driving the burgundy car and the green car. Sometime thereafter, Detective Francis called Ms. Nunnley and told her he was going to arrest the person she had identified. On February 2, 2006, Mr. Minor, who was sixteen years old at the time of the carjacking, was charged with carjacking and three associated offenses.
C.Motion to Admit Expert Testimony
On June 28, 2006, Mr. Minor filed a motion in limine to admit the expert testimony of Dr. Fisher on the reliability of eyewitness identifications. Mr. Minor proffered that “Dr. Fisher would testify about psychological studies that have consistently shown how several factors that are present in this case could have an [e]ffect on the reliability of eyewitness identifications.” Attached to the motion were the results of a 2004 telephone survey of potential jurors in the District of Columbia conducted by Peter D. Hart Research Associates, Inc. (the “Hart survey”) *411that summarized respondents’ answers about the reliability of eyewitness identifications.
The government did not file an opposition to Mr. Minor’s motion to admit the expert testimony of Dr. Fisher nor did the trial court hold an evidentiary hearing. In denying Mr. Minor’s motion, Judge Puig-Lugo looked at nothing more than statistics from the Hart survey and stated that they “squarely address[ed] ... the conclusions that the defense would like to proffer and which underline the conclusion that [six of the seven] matters are not beyond the k[e]n of the average layperson,” the first prong of the Dyas test. For example, Judge Puig-Lugo noted that 80 percent of respondents said it was “false” that “if an eyewitness was under high stress at the time of the crime, the eyewitness will have better recall for the details of the event.” The court concluded that the opportunity for cross-examination and the jury instructions would be sufficient to focus the jury on the factors it should consider in evaluating the reliability of the identification testimony to be given at trial.
D. The Trial
On February 28, 2007, the day trial was scheduled to begin before Judge Craig Is-coe, the prosecution announced that it had just learned that Mr. Kenneth Redfear, a prisoner, claimed he had observed the carjacking while he was panhandling at the 7-Eleven. The court granted a continuance to April 4, 2007, to enable defense counsel time to investigate Mr. Redfear’s story and prepare a defense that would no longer be based on the one-witness case the government had previewed pretrial.
At trial, Mr. Redfear testified that he told the detective that Mr. Minor’s photo “may not be the individual, but this is the closest one of the individuals that’s on this photo array.” He testified that he knew the carjacker because he was the same person who once robbed him and subsequently Mr. Redfear purchased drugs from him. He said the carjacker went by the name of “Snoop” or possibly “Scoop” and that he was a student at H.D. Wood-son High School. The only other evidence presented by the government was the testimony of Ms. Nunnley who said that she “will never forget the face.... I can visualize that gun. And then if I see that gun, I go back to that same moment when I looked directly into his eyes and saw his face.” Mr. Minor also took the stand and testified that at the time of the carjacking he was a student at H.D. Woodson and that “[m]ostly everybody” called him “C.J.,” he had never been known as “Snoop” or “Scoop,” and that he had never robbed or sold drugs to Mr. Redfear. On April 10, 2007, the jury returned a verdict of guilty on all counts. Judge Iscoe sentenced Mr. Minor to concurrent sentences of incarceration of 180 months for armed carjacking, 60 months for armed robbery, 60 months for PFCV, and 24 months for UUV, followed by 5 years of supervised release.
E. Motion for a New Trial
On June 11, 2007, Mr. Minor moved for a new trial based on newly discovered evidence. The defense had learned of a teenager named Kendall Snowden who went by the nickname “Snoop,” bore a striking resemblance to Mr. Minor, and had attended H.D. Woodson High School. According to Mr. Kevann Gardner, one of the defense investigators, Mr. Redfear said that he recognized Mr. Snowden’s picture and said, “had you guys gotten these pictures before trial there’s a good chance your client wouldn’t be in here [in jail] now,” but stated, “I don’t want to get involved. I don’t want to go back to court.” The government opposed Mr. Minor’s motion on the basis of a July 31, *4122007, meeting with Mr. Redfear during which he allegedly “substantially denied the vast majority of the allegations in ... Gardner’s affidavit.”
During the hearing on the motion for a new trial, Mr. Redfear stated that both Mr. Snowden and Mr. Minor were known as “Snoop” and he knew both from the same neighborhood. He also denied telling Mr. Gardner that Mr. Minor likely would not be in jail if he had seen Mr. Snowden’s photo before trial. Judge Iscoe denied the motion, holding that “the ‘newly discovered’ evidence reveals ... nothing more than the discovery of an individual who resembles the Defendant” and noting that the defense’s pre-trial efforts to locate the person called “Snoop ... fell short of diligent.”
F. The Appeal and Remand Order
Mr. Minor appealed, contending that Judge Puig-Lugo’s ruling excluding Dr. Fisher’s expert testimony and Judge Is-coe’s denial of Mr. Minor’s motion for a new trial were each an abuse of discretion. Shortly after oral argument, this court remanded the record to the trial court to conduct a hearing to consider the admission of Dr. Fisher’s proffered testimony in light of Benn II and Russell. We directed the trial court to transmit to this court supplemental findings of fact and conclusions of law addressing the question whether Dr. Fisher’s proffered testimony satisfied the three Dyas factors and, if it concluded that the testimony should be admitted, to grant Mr. Minor a new trial. We reserved judgment on Mr. Minor’s second argument, observing that, if the trial court granted a new trial, Mr. Minor’s argument that the trial court erred in denying his motion for a new trial would be moot.
G. The Proceedings on Remand
On April 11, 2012, Judge Puig-Lugo held a Dyas hearing. Dr. Fisher testified that, among his other qualifications, he had been conducting research on eyewitness memory and identification for more than thirty years and had published more than forty articles on eyewitness identification in “major peer-review[ed] journals” and fifteen to twenty book chapters on the topic.
He then testified about four areas regarding the reliability of eyewitness testimony. First, he explained that “excessive stress impairs later identification, and it also impairs later ability to describe the event that you observed.” Second, he stated that “when a weapon is present the witness is less likely to make a correct identification than when there’s no weapon present,” a phenomenon known as the weapons-foeus effect. Third, with regard to the phenomenon of “exposure-duration,” he testified that “[generally the more time that you have to observe somebody, the more accurate will be your identification.” And finally, with regard to the relationship between a witness’s confidence in her identification and the accuracy of her identification, Dr. Fisher testified that “people tend to over-assess their accuracy.” He acknowledged that confidence is to “some degree predictive of accuracy” but only when assessed at the time of the initial identification. He explained that confidence as a predictor of accuracy generally does not hold true for identifications at trial because witness confidence can be “artificially bolstered” by “confirmatory feedback” as a result of various developments occurring after the witness’s initial identification, such as being told that the person the witness identified is going to be arrested. On cross-examination, Dr. Fisher acknowledged the limits of his testimony by stating that scientists “can never say *413that the witness is right or the witness is wrong ... [t]he best we can do is to indicate to what degree does some factor influence this kind of behavior, and it’s always a probabilistic judgment.”
The government did not call an expert witness. The trial court did not prepare written supplemental findings and conclusions but issued an oral ruling on May 16, 2012. First, the court seemingly held that the proffered expert testimony was beyond the ken of the average juror, thereby satisfying the first Dyas factor.2 However, the court denied Mr. Minor’s motion to admit the expert testimony, finding that the second Dyas prong, which requires that the witness have “sufficient skill, knowledge, or experience in that field or calling as to make it appear that his opinion or inference will probably aid the trier in his search for truth,” had not been met. Dyas, 376 A.2d at 832 (emphasis omitted). The trial court reasoned that because Dr. Fisher “did not present [his conclusions] as absolute truth,” but instead “used qualifiers” such as “might” and “could,” his testimony probably would not aid the trier of fact. Additionally, the court concluded that the testimony would not be helpful because it was based on scientific experiments that did not mirror the “real world.” Finally, the court did not make any findings or conclusions on the third Dyas factor, which would have examined the state of scientific knowledge in the field of expert testimony about the reliability of eyewitness identifications, but, rather, assumed that this factor had been established. Because all three Dyas factors must be satisfied for expert testimony to be admissible, the court’s ruling that Dr. Fisher’s proffered testimony did not satisfy the second prong caused it to again deny Mr. Minor’s motion. The case is now back before this court to determine whether Judge Puig-Lugo properly exercised his discretion in so ruling.
II. DISCUSSION
Over the past several decades, this court has considered the admissibility of expert testimony on the reliability of eyewitness identifications in numerous cases. The social sciences have moved us well beyond the state of our knowledge in Dyas, where, thirty-five years ago, we held, almost as a matter of course, that the trial court did not abuse its discretion in excluding the proffered expert testimony on the reliability of eyewitness identifications. Dyas, 876 A.2d at 831-32. In the intervening years, we have learned much to cause us to reexamine our view that average lay persons serving as jurors are well equipped to call upon their common sense knowledge of the reliability of eyewitness identifications, *414even when aided by cross-examination, to assess the credibility of such testimony. In recognition of these developments, we recently clarified that “Dyas and its progeny do not articulate a per se requirement that all expert testimony about the reliability of eyewitness identification must be excluded.” Russell, 17 A.3d at 586 (quoting Benn II, 978 A.2d at 1277).
To the contrary, our recognition that “eyewitness error is the leading cause of wrongful conviction in the United States,” Benn II, 978 A.2d at 1266 (internal quotation marks omitted), has caused us to impose upon our trial courts a heightened obligation to examine scrupulously a defendant’s proffer of expert testimony in this field. We do not here adopt a new rule, but we reiterate the duty of our trial courts to assess the proffer of expert testimony on the reliability of eyewitness identifications through the sharper lens of the knowledge we have gained. Although the trial courts still retain discretion to admit or exclude expert testimony, “that discretion must be weighed against the constitutional rights of a defendant to present a defense.” Russell, 17 A.3d at 585 (internal quotation marks omitted). Moreover, in exercising that discretion, trial courts “must be guided by the principles that the defense should be free to introduce appropriate expert testimony.” Benn II, 978 A.2d at 1269 (internal quotation marks omitted).
With these developments in mind, we turn to the “two levels of analysis” used in determining whether to admit expert testimony: the “three-fold [Dyas ] test” and the weighing of “probative value ... [versus] ... prejudicial impact.” Ibn-Tamas v. United States, 407 A.2d 626, 632 (D.C.1979); see also Robinson v. United States, 50 A.3d 508, 523 (D.C.2012) (‘Whether expert testimony is helpful to the jury is determined by the well-established criteria ... set forth in Dyas ... and ultimately turns on ‘the relevance and probative value of the proposed scientific evidence’ ”) (internal quotation marks omitted). We review the trial court’s determination for “an abuse of discretion.” Russell, 17 A.3d at 586.
A. The Proffered Testimony Was Beyond the Ken of the Average Juror
In ruling on the first factor, the trial court opined that psychological research on eyewitness identification proffered by Dr. Fisher is “beyond the ken of the average lay person.” This is consistent with our recent observation in Benn II that “[d]espite the fact that jurors may be familiar from their own experience with factors relevant to the reliability of eyewitness observation and identification, ... it cannot be said that psychological studies regarding the accuracy of an identification are within the ken of the typical juror.” Benn II, 978 A.2d at 1277 (alterations in original and internal quotation marks omitted). And while in Dyas we upheld a finding that such studies were not beyond the ken of the average juror, in Benn II we explained that “more recent studies ... confirm” that “jurors generally lack knowledge about psychological studies regarding the accuracy of identifications.” Id. at 1278 n. 90.
Moreover, in Benn II we cited studies assessing jurors’ knowledge on three of the factors affecting eyewitness identifications about which Dr. Fisher was prepared to testify: severe stress, exposure-duration, and the relationship between a witness’s confidence and accuracy. The studies concluded that “jurors believe that the more confident a witness seems, the more accurate that witness’s testimony will be,” but the correlation between a “witness’s expression of certainty in an *415identification and its accuracy is, at a minimum, greatly overstated.” Benn II, 978 A.2d at 1268 & nn. 38-39. With regard to the effect of stress on eyewitness identifications, the studies found that the “average juror is likely to believe that witnesses remember the details of violent events better than nonviolent ones,” but the research shows that the opposite is true. Id. at 1268 & n. 36. As for exposure-duration, the studies found that in violent crimes, “witnesses most often think that the incident lasted longer than it did.” Id. at 1268 & n. 37. In other words, there is less time for the exposure-duration effect to increase the accuracy of an eyewitness’s identification than a lay person might otherwise assume. The scientific studies cited in Benn II are only a small sampling of the research in this field, and they convince us that there is an adequate basis to conclude that the psychological research on eyewitness identification that Dr. Fisher proffered is beyond the ken of the average lay person.3
While finding that the studies were beyond the ken of the average juror, the trial judge also stated that it “[d]oesn’t mean that the conclusions of those articles are beyond the ken of the average lay person,” referring to the 2004 Hart survey.4 The judge was mistaken: it is precisely those conclusions and supporting research, we have recognized, that may be beyond a lay jury’s ken. As our remand order makes clear, citing Benn II, while average jurors may have a “passing familiarity with the potential problems surrounding eyewitness identification,” the conclusions and supporting research drawn from studies in this area are indeed “beyond the ken of most lay persons”; hence, they may be presented through expert testimony citing the findings and underlying research. The determinative question on the need for expert testimony is whether the jurors are “just as competent as the expert to consider and weigh the evidence and draw the necessary conclusions.” Adams v. United States, 502 A.2d 1011, 1021-22 (D.C.1986) (internal quotation marks omitted). Referencing the results of a juror survey, a survey conducted by a market and strategic research firm, does not provide the type of analysis needed to discern whether the average juror is just as competent as an expert like Dr. Fisher to weigh and consider the evidence relating to the reliability of eyewitness identifications. In sum, we have no hesitation in concluding that Mr. Minor’s proffer of Dr. *416Fisher’s expert testimony satisfied the first Dyas factor.5
B. Dr. Fisher Has Sufficient Knowledge Such That His Opinions Will Probably Aid the Trier of Fact
The second Dyas factor requires that the party proffering the expert witness establish that the witness has “sufficient skill, knowledge, or experience in that field or calling as to make it appear that his opinion or inference will probably aid the trier in his search for truth.” Dyas, 376 A.2d at 832 (emphasis omitted). Here, the parties and the trial court agreed that Dr. Fisher’s professional background qualified him to testify as “an expert witness in the field of memory and eyewitness identification.” We see no reason to disturb that conclusion.
There was disagreement, however, as to whether this factor also asks the trial court to assess whether the expert testimony is likely to be helpful to the jury in evaluating the eyewitness identifications, or whether the analysis is limited to determining if the “expert’s credentials [are] sufficient for the type of [expert] testimony proffered.” Ultimately, the trial court adopted the former interpretation and held that Dr. Fisher’s testimony would not probably aid the trier of fact. In doing so, the trial court misconstrued the second Dyas factor. The scope of the second Dyas factor is narrow and assesses only whether the “proffered expert [is] qualified to give” the proposed testimony. Ibn-Tamas, 407 A.2d at 633.
That does not, however, preclude the trial court from assessing the relevance and weighing the probative value versus the prejudicial impact of the proffered expert testimony; indeed, it is required to do so. Robinson, 50 A.3d at 523. Here, the trial court essentially determined that the probative value of Dr. Fisher’s testimony was weak but, because there was no claim of a countervailing danger of unfair prejudice, that was an improper basis upon which to exclude the testimony. Foreman v. United States, 792 A.2d 1043, 1049 (D.C.2002) (noting that the “trial judge has the discretion to exclude relevant evidence” but only when “its probative value is sub*417stantially outweighed by the danger of unfair prejudice”). We address each of the trial court’s reasons for excluding the testimony in turn.
First, the trial judge determined that the testimony should be excluded because the findings of the scientific studies were not absolute and because Dr. Fisher would not say that his theories will always hold true for all people. Pointing to the meta-analysis, a statistical technique for combining and contrasting the findings from independent studies, the trial judge stated, “you have some majority/minority positions” and “[n]ot everybody is on the same page.” For example, the studies found that “severe stress leads to impaired identification” but that “[sjometimes it doesn’t.” The judge was also troubled by Dr. Fisher’s conclusion that the “accuracy of the identification is less likely” when “a weapon is present” due to his use of “qualifying language” and questioned, “when you’re dealing with a lot of mights and a lot of couldn’ts, themes and patterns, where do you cross the line from extrapolation to speculation?”
While we do not reject those observations, the “trial court’s doubts about the certainty of Dr. [Fisher’s] conclusions ... [do] not ... justif[y] excluding the testimony.” Clifford v. United States, 532 A.2d 628, 639 (D.C.1987). Rather, “the degree of certainty with which a particular expert witness proffers an opinion goes to the weight of the testimony,” which is a determination reserved for the jury. Id.; see also Giordano v. Sherwood, 968 A.2d 494, 498 (D.C.2009) (noting that in expert testimony “certainty is not required”; rather, the testimony need only be “based on fact or adequate data”) (internal quotation marks omitted). Although it is true that a trial judge “may exclude outright speculation,” Robinson, 50 A.3d at 523, here the trial court lacked a “firm factual foundation” to find that the conclusions Dr. Fisher discussed in his testimony were tantamount to speculation. Johnson v. United States, 398 A.2d 354, 364 (D.C.1979).
Second, the trial court said it would refuse to let a jury hear Dr. Fisher’s testimony because Dr. Fisher had no direct knowledge of the accuracy of Ms. Nunnley’s identification but rather could describe only “themes” and “patterns” gleaned from scientific studies. In explaining his rationale, the trial court made an analogy: “[W]e might look at all sorts of climatological studies for the month of October in 2011” but that “can only give us themes, patterns”; it “does not tell us whether on October 5, 2011, as we stood outside 500 Indiana Avenue, Northwest and looked at sky, you saw the sun or a cloud.” This caused the court to declare Dr. Fisher’s studies a “mismatch ... as far as the circumstances of this case are concerned.” The analogy is fundamentally flawed because the expert testimony on the reliability of eyewitness testimony was intended to assist the jurors in evaluating Ms. Nunnley’s and Mr. Redfear’s eyewitness identification testimony, and not, as in the example about the weather, to determine conclusively whether their identifications were accurate. Moreover, admission of expert testimony is not contingent on the expert’s having personal knowledge of the witness. See Robinson, 50 A.3d at 527 (“[Testimony could be useful to the jury’s assessment of the witness’s perceptions without the expert evaluating what the actual witness experienced.”); Nixon v. United States, 728 A.2d 582, 592 (D.C.1999) (holding that expert testimony on the battered woman syndrome was admissible even though the “expert ha[d] not examined the complaining witness”). Further, requiring Dr. Fisher to testify about the likelihood that Ms. Nunnley’s and Mr. *418Redfear’s identifications were inaccurate would risk “usurping] the jury’s task of determining witness credibility.” Mindombe v. United States, 795 A.2d 39, 43-44 (D.C.2002) (internal quotation marks omitted).
Third, the trial judge concluded that the testimony would not aid the trier of fact because the studies upon which Dr. Fisher’s testimony relied are “different ... from the real world” and therefore “there’s no ecological validity,” which is “the degree to which the experiment captures the phenomenon that you’re trying to generalize in the real world.” Instead, according to the trial judge, in the studies “[tjhere’s all these factors ... running wild and it varies from case to case, from person to person, from circumstance to circumstance,” but, in contrast, “in the laboratory, seven [out of eight variables] are controlled.” We cannot agree with the trial court’s assessment. Indeed, Dr. Fisher testified that he had no reason to believe that the results of the studies would have been different if the conditions had more closely mirrored the facts of this case. He explained that, because the studies yielded largely “the same results” when conducted in both laboratory and real-world settings, “what you find in the laboratory would hold for the real world.” Boiled down to its essence, the trial court’s concern seems to be that the studies do not provide certainty, a factor we already have stated is within the province of the jury to assess, not the trial judge.
Fourth, in assessing whether the expert testimony would be helpful to the jury, the court repeatedly referred to the Hart survey and stated that the majority of jurors understand the concepts to which Dr. Fisher would testify. But we already have rejected the notion that jurors presumptively know the conclusions drawn from psychological studies of eyewitness identification, such that expert testimony cannot aid them. Neither the trial court nor the government has cited authority for the irrelevance of that testimony just because jurors may have a passing familiarity with the potential problems of this kind of proof.
Beyond the findings discussed above, the trial court did not make any findings regarding the relevance of the proffered expert testimony nor did the court assess the testimony’s probative value versus any prejudicial impact it might have. Nevertheless, as Mr. Minor has argued, there is a reasonable basis for concluding that Dr. Fisher’s testimony is relevant to the facts of this case. “Evidence is relevant if it has any tendency to make the existence of any fact that is of consequence to the determination of the action more or less probable than it would be without the evidence.” Plummer v. United States, 813 A.2d 182, 188 (D.C.2002) (internal quotation marks omitted). Dr. Fisher’s testimony about exposure-duration, extreme stress, and the weapons-focus effect is relevant because Ms. Nunnley testified that the incident lasted only for “seconds” and she was “petrified” when she saw the gun. Further, she testified that “visualizing]” the gun enabled her to remember her assailant’s face. Likewise, the proffered testimony on witness confidence is relevant. Dr. Fisher testified that a witness can become more confident over time, especially when receiving confirmatory feedback about her initial identification, and jurors “rely exclusively, or almost exclusively, on the witnesses] confidence” and “tend to almost disregard all ... other factors.” See also Benn II, 978 A.2d at 1283 (noting that a jury may “automatically infer that an honest profession of certainty ends the inquiry into witness reliability”) (internal quotation marks and footnote omitted). Here, Ms. Nunnley’s first identification lacked a high degree of certainty; she merely pointed to Mr. Minor’s photo and said, “[t]hat looks like him,” without expía-*419nation. See Benn v. United States (Benn I), 801 A.2d 132, 145 (D.C.2002) (noting that saying a person “looks like” the assailant is “not really an identification at all” because “[c]ommon sense tells us that many people resemble one another”). Yet later, Ms. Nunnley received confirmatory feedback on her identification from Detective Francis when he told her that he was going to arrest the person she identified as the culprit, and at trial she was very confident in her identification of Mr. Minor.
Thus, it is clear that Mr. Minor’s “proffers contained sufficient detail to signify that if the jurors credited the expert’s testimony, it would assist them” in assessing the testimony of the eyewitnesses. Robinson, 50 A.3d at 525. Therefore, “[o]n this record, we are comfortable concluding that the expert’s testimony would not have left the jury open to speculate and that the trial court erred in imposing an unduly stringent standard of relevance.” Id.6
C. The State of the Scientific Knowledge Permitted Dr. Fisher to Assert a Reasonable Opinion
The third Dyas factor permits a trial court to exclude expert testimony if the “state of the pertinent art or scientific knowledge does not permit a reasonable opinion to be asserted even by an expert.” Dyas, 376 A.2d at 832. When the expert testimony is based on “a novel scientific test or a unique controversial methodology or technique,” a so-called Frye7 test must be conducted, (Ricardo) Jones v. United States, 27 A.3d 1130, 1137 (D.C.2011) (internal quotation marks omitted), “under which scientific testimony is admissible only if the theory or methodology on which it is based has gained general acceptance in the relevant scientific community.” (John) Jones v. United States, 990 A.2d 970, 977 (D.C.2010) (citing Frye v. U.S., 293 F. 1013 (D.C.Cir.1923)). “Once a technique has gained such general acceptance, we will accept it as presumptively reliable and thus generally admissible into evidence.” (Ricardo) Jones, 27 A.3d at 1136 (internal quotation marks omitted).8
The trial court neither determined whether a Frye hearing was necessary nor decided whether the third Dyas factor had been satisfied, stating that “even if we assume ... that [the] third prong of Dyas has been established .... the second *420prong has not been met.” Nevertheless, although we agree with Mr. Minor that a full Frye hearing was not required because “social science methodology regarding the strength and weaknesses of eyewitness identification evidence is [not] ‘novel’ or ‘experimental,’ ” we provide a basis for our conclusion that the underlying methodology has gained general acceptance in the relevant scientific community. Cf. Pettus v. United States, 37 A.3d 213, 217 (D.C.2012) (concluding that it was proper for this court to determine whether “handwriting identification meets Frye’s general acceptance standard” where expert testimony on the topic arguably was not novel but where such a determination had not previously been announced in our cases).
In Benn II we observed that it “can credibly be argued” that research on the reliability of eyewitness testimony “has reached that critical juncture” from a “theory, initially untested, unrecognized, and unsupported by evidence” to one that now “receive[s] widespread recognition and the support of experts in the ... field” and noted that “[w]hereas once we could only speculate as to the inaccuracy of an eyewitness identification, now there is published scientific research that questions its accuracy when made under certain conditions.” Benn II, 978 A.2d at 1278. This is consistent with Dr. Fisher’s testimony that there is a “consensus in the scientific community that the methodology that underlies all of the studies [that he drew upon in his testimony], including the meta-analysis, is sound[.]” The studies relied upon by Dr. Fisher were published in major peer-reviewed journals, which only accept articles where the underlying “methodology ... conform[s] to generally accepted [research] principles.”
To support a determination that the methodology underlying the eyewitness identification research does not pass muster under the Frye analysis, there must be “scientists significant either in number or experience [that] public[ly] oppose” it. United States v. Jenkins, 887 A.2d 1013, 1022 (D.C.2005). We find no evidence of such a disagreement here. To the contrary, statistics cited in the government’s brief support our conclusion. In the 2001 survey of experts,9 “60% of the experts surveyed agreed that the impact of [very high levels of] stress on the accuracy of eyewitness testimony was a reliable topic on which to testify.” In fact, the experts surveyed overwhelmingly said that the factors affecting eyewitness identification about which Mr. Minor sought to have Dr. Fisher testify were reliable enough for expert testimony.10
In its opposition to Mr. Minor’s motion to admit expert testimony, the government argued that a reasonable opinion cannot be asserted by an expert because “psychologists continue to disagree on the results of studies testing eyewitness memory and perception.” (Emphasis added.) The government focused on the wrong question. The issue is not “the acceptance of a particular ... conclusion derived from [the] *421methodology”; rather, the question is the acceptance of the methodology itself. Jenkins, 887 A.2d at 1022; Burgess v. United States, 958 A.2d 1055, 1063 n. 12 (D.C.2008) (“In summary, satisfaction of the third Dyas criterion begins — and ends— with a determination of whether there is general acceptance of a particular scientific methodology, not an acceptance, beyond that, of particular study results based on that methodology.”) (quoting Ibn-Tamas, 407 A.2d at 638).11 Accordingly, we are satisfied that Dr. Fisher’s proffered testimony satisfies the third Dyas prong be*422cause the theory or methodology on which it is based has gained broad general acceptance in the relevant scientific community.
D. The Exclusion of Dr. Fisher’s Testimony Was Not Harmless Error
Finding the exclusion of Dr. Fisher’s testimony was in error, we must now assess whether the error was harmless. Whether or not the “trial court[’s] evidentiary error in ... denying expert opinion testimony is harmless is governed in this case by the non-constitutional standard announced” in Kotteakos v. United States, 328 U.S. 750, 765-66, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946). See Smith v. United States, 27 A.3d 1189, 1196 (D.C.2011). An error is not harmless if we “cannot say, with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the error.” Kotteakos, 328 U.S. at 765, 66 S.Ct. 1239. In our assessment, we “take into account factors such as the closeness of the case, the centrality of the issue affected by the error, and any steps taken to mitigate the error, but our focus is on the likely impact of the alleged error on the jury’s verdict.” Robinson, 50 A.3d at 528 (internal quotation marks omitted). We conclude that there is a reasonable probability that Dr. Fisher’s testimony would have altered the jury’s verdict in Mr. Minor’s favor.
Here, the government’s only evidence was the identification testimony of Ms. Nunnley and Mr. Redfear. As in Russell and in Benn II, where we found that the exclusion of expert testimony on the reliability of eyewitness identifications was not harmless, “[t]here was no scientific evidence linking [Mr. Minor] to the crime[,][t]here was no evidence that [he] was apprehended with the weapon or proceeds of the crime, and he made no statements implicating himself.” Russell, 17 A.3d at 589; see also Benn II, 978 A.2d at 1264. Precluding the admission of expert testimony on the reliability of eyewitness identification in a case “grounded on eyewitness identifications of a stranger,” where there is little “other corroborating evidence,” and where “the defense depends entirely upon demonstrating that the identifying witnesses are not as reliable as they believe themselves to be,” is not harmless “provided that the facts underlying the identification establish a sound foundation for applying the principles expounded by the expert.” Benn II, 978 A.2d at 1283.
As we discussed earlier, the principles Dr. Fisher would have testified about are germane to these eyewitness identifications. Thus, Dr. Fisher’s testimony would have provided a basis for Mr. Minor to cast doubt on the reliability of Ms. Nunn-ley’s and Mr. Redfear’s identifications. The government argues that “any error was harmless because appellant fully and forcefully presented his misidentification defense.” This argument misses the point that, “[without the expert’s testimony, appellant had no factual underpinnings for the scientific theories ... that might cast doubt on the eyewitness’ testimony.” Russell, 17 A.3d at 589.
The government further argues that any error was harmless because the identification by Ms. Nunnley, a stranger, was corroborated by an identification by a “non-stranger,” i.e., Mr. Redfear. The government analogizes this case to Heath v. United States, 26 A.3d 266 (D.C.2011), in which this court concluded that exclusion of expert testimony on the reliability of eyewitness identifications was not prejudicial to the appellant where there were “mutually reinforcing” identifications made by strangers and someone who knew the *423defendant. Id. at 284. But Heath is distinguishable from the circumstances here.
First, in Heath, there were not one but two non-stranger eyewitnesses, whereas here there was only one. 26 A.3d at 282. Second, the appellant in Heath offered “no explanation, other than sheer coincidence,” for why one of the non-stranger witnesses would have misidentified the appellant as the perpetrator. Id. at 284. Here, Mr. Minor proffered numerous reasons why Mr. Redfear would have incorrectly identified him as the earjacker. As an initial matter, Mr. Redfear’s motivation to curry favor with the government can scarcely be doubted, given his lengthy criminal record and the fact that he was incarcerated, serving sentences on convictions of UUV and bail jumping and had just been arrested for burglary at the time he made his identification. Moreover, Mr. Redfear’s identification was heavily qualified: Upon being shown a photo array, he said, “this may not be the individual” but nonetheless selected Mr. Minor’s photo as “the closest one ... on this photo array.” In fact, as we recognized in discussing Ms. Nunnley’s identification of Mr. Minor, Mr. Redfear’s identification was “not really an identification at all” because “[cjommon sense tells us that many people resemble one another.” Benn I, 801 A.2d at 145. And the uncertainty of Mr. Redfear’s selection of Mr. Minor from the photo array was not ameliorated by his in-court identification; as we have said before, identifications during trial are “by no means ... overwhelming” because the courtroom is an inherently “suggestive setting.” Id.
The weaknesses in Mr. Redfear’s identification do not stop there. Mr. Redfear testified that he knew Mr. Minor as “Snoop,” but there is no evidence that Mr. Minor had ever been known as “Snoop.” Further, Mr. Redfear testified that he saw Mr. Minor, “the same Snoop who did the carjacking,” out after midnight during a time when Mr. Minor was at home under an electronically monitored curfew after 8 p.m., and Mr. Minor’s case manager confirmed there was no record of a curfew violation. Given all these weaknesses in Mr. Redfear’s testimony, we agree with Mr. Minor that Mr. Redfear’s identification did not provide corroborating evidence nearly as strong as the testimony of the non-stranger witnesses in Heath.12
In sum, the government’s case against Mr. Minor certainly was not “strong or overwhelming,” Smith, 27 A.3d at 1198, and we conclude that “[ejxpert testimony concerning the [reliability of eyewitness identifications] might well have created a reasonable doubt in the minds of jurors.” Id. at 1199. Consequently, we “cannot say[ ] with fair assurance ... that the judgment was not substantially swayed by the error.” Kotteakos, 328 U.S. at 765, 66 S.Ct. 1239.
III. CONCLUSION
We reverse the judgment convicting Mr. Minor of the February 2, 2006, carjacking and the offenses associated therewith and remand for a new trial at which Mr. Minor shall be permitted to introduce Dr. Fish*424er’s expert testimony on the reliability of eyewitness identifications.13
Reversed and remanded.
. The offenses of conviction are covered under the following statutes: carjacking while aimed, D.C.Code §§ 22-2803, -4502 (2001); armed robbery, D.C.Code §§ 22-2801, -4502; PFCV, D.C.Code § 22-4504(b); and UUV, D.C.Code § 22-3215.
. We say "seemingly” because the trial court’s oral ruling is not entirely clear:
I understand that studies that are published in journals are ... beyond the ken of the average lay person unless they sit around at home reading these journals, which is probably highly unlikely. Doesn’t mean that the conclusions of those articles are beyond the ken of the average lay person.
* }¡« í¡í !)s s}c ‡
Ultimately, the issue is whether people know what those articles conclude. And again, the articles might not be within the k[e]n of the average lay person. The studies might be beyond the ken of the average lay person. But it's clear from [the Hart] survey that the average lay person understands the concerns regarding identification.
We understand the trial court to be acknowledging the distinction drawn in our decision in Benn II between jurors' experiential familiarity with factors relevant to the reliability of eyewitness observations, on the one hand, and psychological studies that explain the scientific bases underlying the experiential familiarity, on the other hand. Benn II, 978 A.2d at 1277.
. We leave for determination on retrial the question whether or not the so-called weapons-focus effect is beyond the ken of the average juror. At the hearing, Mr. Minor’s counsel asserted that the weapons-focus effect is “part of stress” and therefore the court should find that it is also beyond the ken of the average juror. However, Dr. Fisher could not confirm that assertion with a cite to the literature, although he believed it should be true. Accordingly, we cannot extrapolate from the findings of studies that show that the effect of highly stressful situations on identifications is beyond the ken of the average lay person to the weapons-focus effect. We note, however, that a 2001 survey of 64 "eyewitness experts,” a survey submitted to the trial court by the government, found that only thirty-four percent of the experts believed that the weapons-focus effect is a “matter of common sense” that "most jurors believe.” Saul M. Kassin, V. Anne Tubb, Harmon M. Hosch, & Amina Memon, On the “General Acceptance” of Eyewitness Testimony Research, 56 Am. Psychologist 405, 407-08, 412 (2001).
. This is the same survey on which the court based its initial decision to exclude Dr. Fisher's testimony. As we stated in our remand order, reliance on the survey alone did not constitute "the careful consideration that we now require [in light of Benn II and Russell ] before the trial court excludes expert testimony on the reliability of eyewitness identification.”
. We are not alone in holding that the topics about which Dr. Fisher would testify are beyond the ken of the average juror. See, e.g., United States v. Brownlee, 454 F.3d 131, 142 (3d Cir.2006) (noting that "jurors seldom enter a courtroom with the knowledge that eyewitness identifications are unreliable” and "while science has firmly established the inherent unreliability of human perception and memory, this reality is outside the jury's common knowledge, and often contradicts jurors’ commonsense understandings”) (internal quotations marks and citations omitted); United States v. Moore, 786 F.2d 1308, 1312 (5th Cir.1986) (stating that "the conclusions of the psychological studies are largely counter-intuitive, and serve to explode common myths about an individuals capacity for perception”) (internal quotation marks omitted); State v. Guilbert, 306 Conn. 218, 49 A.3d 705, 720-21 (2012) (noting "widespread judicial recognition that eyewitness identifications are potentially unreliable in a variety of ways unknown to the average juror” which "tracks a near perfect scientific consensus”); State v. Hill, 463 N.W.2d 674, 677 (S.D.1990) ("[E]ven though it is safe to say that jury members have some experience and common sense knowledge of factors that may cause occasional mistakes in identification, they do not possess the expert’s comprehensive training in assessing the reliability of identification.”); People v. McDonald, 37 Cal.3d 351, 208 Cal.Rptr. 236, 690 P.2d 709, 721 (1984) (concluding that "although jurors may not be totally unaware of the foregoing psychological factors bearing on eyewitness identification, the body of information now available on these matters is sufficiently beyond common experience that in appropriate cases expert opinion thereon could at least assist the trier of fact”) (internal quotation marks omitted), overruled on other grounds by People v. Mendoza, 23 Cal.4th 896, 98 Cal.Rptr.2d 431, 4 P.3d 265 (2000).
. Our holding does not, of course, preclude a trial court from determining in an appropriate case that particular expert testimony, even if it satisfies all three Dyas factors, should be excluded as more prejudicial than probative on the facts of the particular case. For example, in a non-stranger identification case a trial court could rule that the relevance of the expert's views might be too slight compared to the likelihood of distracting or confusing the jury. We hold only that no such determination could have been made in this case.
. Frye v. United States, 293 F. 1013 (D.C.Cir.1923).
. We reject the government’s various positions on the third prong of the Dyas test. It first argued in opposition to Mr. Minor’s motion in limine on remand that "the state of the ... scientific knowledge does not permit a reasonable opinion to be asserted even by an expert,” a position we find untenable in light of Benn II, and later softened its position to contend only that it was willing to assume "that prong three was satisfied for the purpose of this hearing. The government is not conceding that ... [w]e’re assuming it for the purposes of this hearing." As Mr. Minor correctly explained in argument on remand, the third Dyas factor is not something that varies from case to case or courtroom to courtroom; once a particular field of science has reached a state of general acceptance, it is presumptively admissible although the "party opposing the evidence, of course, may challenge the weight the jury ought to give it.” (Ricardo) Jones, 27 A.3d at 1136 (internal quotation marks omitted).
. See supra note 3.
. The survey found that eighty-seven percent of experts agreed that it is reliable to testify that "presence of a weapon impairs an eyewitness's ability to accurately identify the perpetrator’s face”; eighty-one percent said the phenomenon that the "less time an eyewitness has to observe an event, the less well he or she will remember it” is reliable for expert testimony; and eighty-seven percent indicated that it is reliable to testify that "eyewitness’s confidence is not a good predicator of his or her identification accuracy,” and relatedly, ninety-five percent said the phenomenon that "eyewitness’s confidence can be influenced by factors that are unrelated to identification accuracy” is reliable for expert testimony. Kassin, supra note 3, at 408, 412.
. Beyond our decision in Benn II, further assurance that the underlying methodologies have gained general acceptance is that courts in numerous other jurisdictions have held that they are generally accepted or are sufficiently reliable to permit a qualified expert to testify about the reliability of eyewitness identifications. See, e.g., United States v. Moore, 786 F.2d 1308, 1312 (5th Cir.1986) ("This Court accepts the modern conclusion that the admission of expert testimony regarding eyewitness identifications is proper[ ].... We cannot say such scientific data is inadequate or contradictory. The scientific validity of the studies confirming the many weaknesses of eyewitness identification cannot be seriously questioned at this point.") (internal quotation marks omitted); United States v. (James Darnell) Smith, 736 F.2d 1103, 1107 (6th Cir.1984) ("The day may have arrived, therefore, when testimony [on the reliability of eyewitness identifications] can be said to conform to a generally accepted explanatory theory.”); United States v. Smith, 621 F.Supp.2d 1207, 1212-13 (M.D.Ala.2009) (concluding that because the expert’s "methods satisfied the reliability prong of Daubert [v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993)]" and the "theories underlying [his] testimony have been well-tested in peer-reviewed publications,” the expert was "equipped to provide the jury with reliable expert testimony on eyewitness identifications"); State v. Guilbert, 306 Conn. 218, 49 A.3d 705, 721, 732 (2012) (”[T]he scientific evidence” regarding the reliability of eyewitness identifications "is both reliable and useful” as the "[experimental methods and findings have been tested and retested, subjected to scientific scrutiny through peer-reviewed journals, evaluated through the lens of meta-analyses, and replicated at times in real-world settings” and "competent expert testimony predicated on those studies’ findings satisfies the threshold admissibility requirement") (first alteration in original and internal quotation marks omitted); State v. Henderson, 208 N.J. 208, 27 A.3d 872, 916-17 (2011) (adopting the findings of the special master that the "science abundantly demonstrates the many vagaries of memory encoding, storage, and retrieval; the malleability of memory; the contaminating effects of extrinsic information ... and the many other factors that bear on the reliability of eyewitness identifications” and recognizing that these “estimator variables [such as stress, weapons focus, duration, and level of certainty demonstrated at the confrontation] can affect the reliability of eyewitness identifications”); State v. Clopten, 223 P.3d 1103, 1112 (Utah 2009) (overturning the court’s de facto presumption against admissibility of eyewitness expert testimony and predicting that its holding "will result in the liberal and routine admission of eyewitness expert testimony, particularly in cases where, as here, eyewitnesses are identifying a defendant not well known to them”); People v. LeGrand, 8 N.Y.3d 449, 835 N.Y.S.2d 523, 867 N.E.2d 374, 377 (2007) (stating that research on the "reliability of eyewitness identifications ... [is] produced through sound, generally accepted experimentation techniques and theories, published in scholarly journals and subjected to peer review [and] have over the years gained acceptance within the scientific community"); State v. Copeland, 226 S.W.3d 287, 299 (Tenn.2007) (noting that "[t]imes have changed” and "[tjoday, many scholarly articles detail the extensive amount of behavioral science research” on the reliability of eyewitness identifications and it is the "empirical science behind the reliability of eyewitness testimony that persuades” the court to hold that such testimony is admissible); Jules Epstein, The Great Engine That Couldn’t: -Science, Mistaken Identifications, and the Limits of Cross-Examination, 36 Stetson L.Rev. 727, 740-41 (2007) (The "Kassin [study's] accumulation of studies and search for consensus make clear that, whether applying the Frye or Daubert standard for admitting expert testimony, the methodology and conclusions of these experts are sufficiently reliable to be admissible in court.”).
. The weaknesses in Mr. Redfear's identification of Mr. Minor were apparent not only in his trial testimony but also in the evidence proffered in support of Mr. Minor’s motion for a new trial. For example, Mr. Minor offered evidence that seventeen-year-old Kendall Snowden, who looks similar to Mr. Minor and had attended the same high school as Mr. Minor, went by the name of "Snoop.” After being confronted with a photo of Mr. Snow-den, Mr. Redfear claimed that both Mr. Snowden and Mr. Minor were known as "Snoop” and he knew both from the same neighborhood, which, we agree with appellant, is a questionable coincidence at best.
. In light of our holding, we of course have no need to address Mr. Minor's argument that the trial court erred in denying his motion for a new trial.