After a jury trial, Todd Matthew Thomas was convicted of five counts of first-degree burglary, in violation of D.C.Code § 22-801(a) (2007 Supp.) (one count each from the B.C., C.T., F.H., W.C., and M.C. incidents); one count of attempted second-degree burglary (the S.C. incident), in violation of D.C.Code §§ 22-801(b), -1808 (2007 Supp.); two counts of assault (one count from the W.C. incident and one count from the M.C. incident), in violation of D.C.Code § 22-404 (2007 Supp.); and one count of fourth-degree sexual abuse (from the B.C. incident) with aggravating circumstances, in violation of D.C.Code §§ 22-3005(2), -8020(a)(5) (2007 Supp.).1 On appeal, Thomas contends the trial court erred by (1) admitting evidence of his prior sexual assault conviction in Virginia pursuant to Drew v. United States, 118 U.S.App. D.C. 11, 331 F.2d 85 (1964), (2) excluding his proffered Winfield2 defense, and (3) excluding his expert’s testimony about factors that might have affected the eyewitnesses’ identifications of him. Holding the trial judge erred by admitting the other-crimes evidence, we reverse the convictions and remand the case for a new trial.
I.
The government alleged that on six separate occasions between July 2007 and August 2008, Thomas entered or attempted to enter the homes of male Georgetown University students and, in some instances, assaulted or attempted to assault them as they slept. Aside from one assault, which occurred at 1320 35th Street, N.W., the remaining assaults all occurred in two adjoining townhomes, located at 1207 and 1209 33rd Street, N.W. The 35th Street townhouse is approximately three blocks from the 33rd Street townhouses in the Georgetown section of the District of Columbia.
At trial, B.C., a Georgetown University student, testified that on July 28, 2007, Thomas sexually assaulted him, while he was sleeping in his upstairs bedroom at 1207 33rd Street, N.W. Before the assault, B.C. had six or more alcoholic beverages at a bar, and returned home shortly after 2:00 a.m. B.C. testified that he fell asleep, awoke and found Thomas sitting on the edge of his bed, “touching [him] sexually.” According to B.C., Thomas had his hand on his penis, and was “jerking [him] off.” As B.C. started “to become more awake,” Thomas said, “[n]o, I’m leaving, I’m leaving,” and he left. Although B.C. was “alarmed and confused,” he “eventually fell back asleep.” He did not call the police to report the incident.
C.T., another Georgetown University student, testified that in September of 2007, while he was sleeping, Thomas entered his bedroom at 1320 35th Street, N.W. Prior to this incident, C.T. and his roommates hosted a party at their house, and C.T. had several drinks. C.T. testified that the house was “completely empty” around 2:30 a.m., and he had not locked the front door. According to C.T., Thomas *1256told C.T. that he was there for the party and that he was looking for his girlfriend. C.T. told Thomas that his girlfriend was not there and that he should leave. C.T. then “forcefully” escorted Thomas out of the house. C.T. testified that he saw Thomas again, approximately seven to eight months later, in the summer of 2008, sitting in the driver’s seat of a champagne-colored Lexus SUV. C.T. noted a partial license plate number of the vehicle, and called a police officer friend, David Pritch-ett, and gave him the partial license plate numbers, along with the make and model of the SUV.3
F.H., a Georgetown University student,, testified that he woke up on June 8, 2008 at sometime after 4:00 a.m., to find Thomas sitting on the side of his bed. F.H.’s bed was located in the same bedroom, at 1207 33rd Street, N.W., where B.C. had been assaulted the year before. On the night at issue, F.H. and his roommates had hosted a party. The house had “cleared out” by 1:30 a.m., and F.H. went to bed shortly after that. According to F.H., four or five hours later, he woke up to find Thomas sitting on the side of his bed. F.H. told Thomas to leave four or five times, and Thomas “acted like” he was going to leave. F.H. claimed that Thomas went down the steps, but then came back up. Thomas then walked down the hallway toward a bedroom where F.H.’s roommates were sleeping, and then returned to F.H.’s room. At that point, F.H. began screaming at Thomas to leave the house, and Thomas promptly left. F.H. reported the incident to the police that morning.
W.C., a Georgetown University student who lived at 1209 33rd Street, N.W., testified that he woke up on June 22, 2008 to Thomas massaging his shoulders in his living room. Prior to the incident, W.C. had drunk approximately twelve beers, and between 2:00 a.m. and 3:00 a.m., had fallen asleep on the couch in his living room. According to W.C., he then woke up to find Thomas massaging his shoulders. W.C. stood up, told Thomas “[g]et off me,” and went into the bathroom to remove his contact lenses. W.C. testified that he then saw Thomas in the backyard, trying to enter 1207 33rd Street, N.W. (the attempted second-degree burglary charge). After explaining to Thomas that he could not leave that way, he led Thomas through the house at 1209 33rd Street, N.W., and out of the front door. W.C. did not report the incident to the police.
M.C., another Georgetown University student, testified that he woke up in August 2008 to Thomas massaging his ankles. M.C. had several drinks that night, and fell asleep on the couch in the living room of 1209 33rd Street, N.W., around 2:15 a.m. According to M.C., at approximately 4:30 a.m., he woke up to Thomas sitting on the arm of the couch, massaging his ankles. M.C. told Thomas to “get out,” and Thomas tried to calm M.C. down. M.C. continued telling Thomas to leave, and Thomas left the house. M.C. admitted that Thomas may have apologized, and that he gave him a “high-five or ... low-five” before he left. M.C. then spoke with his roommate, P.H., and after giving P.H. a description of the assailant, he called the police.4
*1257That same morning, between 5:30 a.m. and 6:00 a.m., Officer James Culp stopped Thomas because his heavily intoxicated passenger was hanging out of the vehicle. According to Officer Culp, a lookout description of the burglary suspect in the M.C. incident was given while Thomas was stopped. Officer Culp thought Thomas matched the description of the lookout, and called Officer Pritchett to let him know that he had a suspect stopped who matched the description. Thomas was asked to drive his vehicle to the front of 1209 33rd Street, N.W. so they could conduct a show-up identification. While Thomas was at the scene of the burglary, both M.C. and W.C. identified him as their assailant. M.C. was present when W.C. identified Thomas, and W.C. was present when M.C. identified Thomas. Officer Pritchett compared the license plate number on Thomas’s vehicle to the license plate number that he had received from C.T., and determined that they were the same. Officer Pritchett called C.T., and told him that he “need[ed] to come down” because he thought they “ha[d] the guy [C.T. was] looking for.” C.T. walked to the scene, and identified Thomas as his assailant (in the presence of W.C. and M.C.).
Later that morning, Detective Andrew Way conducted an interview of Thomas during which Thomas admitted that he had entered the house at 1209 33rd Street, N.W., assertedly in search of a party. Thomas said that he tapped M.C.’s shoulder while he was sleeping on the couch, and M.C. informed him the party was over, they shook hands, and Thomas left. Thomas denied fondling M.C.’s legs, and claimed that it was a “[misunderstanding.”
Prior to trial, B.C. and F.H. both selected a photo of Thomas out of a nine-person photo array, and B.C. identified Thomas in court as the person who sexually assaulted him. F.H. also testified that he spoke to his roommate, P.H., about his assailant’s description before picking Thomas from the photo array.5 F.H. related that Thomas was in his room in June 2009, but he did not make a photo identification until August 2009.
At trial, M.C. and W.C. each identified Thomas in court, and testified that they were “100 percent” sure of their on-scene identifications. C.T. also identified Thomas in court, and testified that he had “no doubt” that Thomas was the same person he saw in his room and identified at the burglary scene. F.H. testified that he was “fairly certain” Thomas was his assailant.
In addition, the government presented evidence that Thomas had committed a sexual assault in Virginia. Through the testimony of the victim in that assault, the jury heard evidence that Thomas had brought the victim, who was intoxicated, to Thomas’s apartment in Virginia on December 31, 2008. Thomas sexually assaulted the victim, who reported the incident to the police the following day.
In his defense, Thomas sought to introduce Winfield evidence that another individual, described in the media as the “Georgetown Cuddler,” who sexually assaulted female Georgetown University students about the time that Thomas was being electronically monitored on these charges,6 also committed the crimes that Thomas was charged with. However, the *1258trial court precluded Thomas from introducing that evidence, finding Thomas had failed to proffer a sufficient nexus between the cited incidents and the offenses charged in this case:
The problem I have with the — the instances that you proffer in support of your Winfield motion is that they all involve women as victims. And there’s no offense charged with respect to Mr. Thomas that involves women at all. And the one incident that you charge or that you cite to where a man is involved, the suspect in that instance entered the house, apparently saw the man, and then left. I don’t think ... that’s enough to make any connection between that conduct and the conduct in this case charged to Mr. Thomas.
The trial court concluded that because the proffered incidents were “dissimilar in terms of the conduct in the victims,” the evidence “would be distracting to the jury” and would have had little or no probative value.
Thomas also sought to present expert testimony on identification regarding, among other things, post-event contamination and the correlation between confidence and accuracy. The trial court precluded Thomas’s expert from testifying because “under the facts and circumstances” the expert’s testimony would not be “beyond the keen [sic] of the average juror” nor would it “aid the trier of fact in this case.” Subsequently, Thomas sought reconsideration of the trial court’s ruling, and provided it the Kassin study, Saul M. Kassin et al., On the “General Acceptance of Eyewitness Testimony Research: A New Survey of the Experts, 56 Amer. Psychol. 405 (2001). However, the trial court again concluded that the expert’s testimony would not aid the trier of fact because post-event contamination and the correlation between confidence and accuracy, among other things, were not beyond the ken of the ordinary juror.
During his closing argument, the prosecutor made the following argument regarding the witnesses’ confidence in their identifications:
There are several theories that may be advanced to explain what happened in this case. One of [t]hose theories is misidentification. Ladies and gentlemen, you have met six victims in this case. You’ve met [C.T.]. You’ve met [W.C.]. You’ve met [M.C.]. All three of those individuals were present on August 22nd. All of them were certain that it was the defendant who’d been sitting there on [C.T.j’s bed, who had been sitting there when [W.C.] was sleeping on the couch, who was massaging [M.C.]’s ankles. All of them are certain of that.
You also have identifications from [B.C.]. [B.C.] picked out the defendant. [R.L.] picked the defendant and picked this individual in position No. 9. And [F.H.] also picked the defendant. It is no accident that when viewing this photo array, [R.L.], [B.C.], and [F.H.] picked the defendant. And that happens to be the same person that [M.C.] and [W.C.] and that [C.T.] also positively identified.
II.
Thomas first argues the trial court erred by admitting evidence relating to his previous conviction in Virginia for sexual assault. In that case, Thomas pled guilty to aggravated sexual battery based on a December 31, 2008 incident involving a victim in Arlington, Virginia.7 The Arlington victim testified about that crime at *1259Thomas’s trial in this case. Specifically, he testified that he worked as a waiter at the Liberty Tavern in Arlington, Virginia. On New Year’s Eve, he drank about “six glasses of wine slash champagne,” and he could “barely remember trying to hail a cab.” The next thing the victim remembered was waking up, at approximately 5:30 a.m., in Thomas’s apartment with his pants down and Thomas fondling his penis. The victim “black[ed] out” and woke up in a downstairs apartment in the same building around 11:00 a.m. He went home and reported the incident to the Arlington County Police. The government referenced the Virginia incident in its closing argument, saying
[y]ou also have the benefit of additional evidence, evidence of the defendant’s motive, absence of that this was all some big mistake or accident, evidence of his identity. And this evidence is actually the testimony of [the victim]. [The victim] was intoxicated New Year’s Eve, and he woke up to the defendant masturbating his penis without his consent, without his permission. Is that also all one big misunderstanding?
The trial court gave the jury a limiting instruction informing the jury that it could only consider this other-crimes evidence for the purpose of proving “motive, identity, common scheme or plan, [and] the absence of mistake or accident.” We conclude that the trial judge abused discretion in permitting that evidence to be presented.
“A decision on the admissibility of evidence, of course, is committed to the sound discretion of the trial court, and we will not disturb its ruling absent an abuse of discretion.” Sanders v. United States, 809 A.2d 584, 590 (D.C.2002) (quotation marks omitted). Moreover, “the evaluation and weighing of evidence for relevance and potential prejudice is quintessentially a discretionary function of the trial court, and we owe a great degree of deference to its decision.” Johnson v. United States, 683 A.2d 1087, 1095 (D.C.1996).
The lead case on other crimes evidence is Drew,8 which we have said holds that
[e]vidence of crimes, independent of the crime charged, are inadmissible to prove a defendant’s disposition to commit the crime charged. However, such evidence is admissible for legitimate purposes, such as to prove motive, intent, absence of mistake or accident, a common scheme or plan, or identity of the person charged with the crime on trial.
Frye v. United States, 926 A.2d 1085, 1091-92 (D.C.2005) (internal citations omitted).
Furthermore,
[t]he Drew exceptions for intent, motive, and absence of mistake are applicable only when the defendant’s state of mind is a material or genuine issue in the case.... The defendant’s state of mind becomes an issue when he makes it an issue by raising an affirmative defense ... or when he has been accused of a specific intent crime.
Howard v. United States, 663 A.2d 524, 528 n. 6 (D.C.1995) (internal citations and emphasis omitted).
As we stated in Newman v. United States, “other crimes evidence offered by the prosecution requires for admission a fairly rigid standard of similarity[] to be sure the defendant is not convicted for mere propensity to commit crime[.]” 705 A.2d 246, 255 (D.C.1997) (quotation marks omitted). “[T]here must be enough points of similarity in the combination of circumstances surrounding the two crimes,” Ifelowo v. United States, 778 A.2d 285, 290 *1260(D.C.2001) (quoted language refers to discussion of prejudicial joinder) (quotation marks omitted), to “ ‘demonstrate that there is a reasonable probability that the same person committed both due to the concurrence of unusual and distinctive facts relating to the manner in which the crimes were committed.’ ” Brooks v. United States, 448 A.2d 258, 257 (D.C.1982) (quoting Drew, supra, 118 U.S.App.D.C. at 16, 331 F.2d at 90). We “consider the totality of the factual circumstances” in determining whether there is “a sufficient basis for admission under the Drew doctrine.” Easton v. United States, 533 A.2d 904, 907 (D.C.1987) (quotation marks and emphasis omitted).
Here, the government defends the trial court’s rulings on three grounds: (1) that the evidence was admissible to show Thomas’s motive in committing the charged offenses; (2) that it was relevant to show Thomas’s identity “with regard to the sexual assault of [B. C.]”; and (3) that the evidence was relevant to show the absence of mistake and state of mind “with respect to” the charges “involving [M. C.].”9 Each of the government’s arguments fail for the reasons set forth below.
First, the government argues that the evidence of the aggravated sexual assault conviction was admissible because it showed “that [Thomas] was motivated to prey on intoxicated young white men in vulnerable positions.” This court expressly rejected a similar argument in Harrison v. United States, 30 A.3d 169 (D.C.2011). There, the appellant was convicted of, among other things, child sexual abuse stemming from an allegation that he kissed and groped a tenth-grade female student, whom he met while serving as a youth mentor. Id. at 171-73. The trial court allowed the government to introduce, under the motive exception, evidence by “three other female students at the high school that Harrison had made sexually suggestive comments to them prior to his alleged sexual assault on [the victim].” Id. at 171. The only specific motive for assaulting the victim the government argued was an “attraction] to teenage girls.” Id. at 179.
We reversed, holding that the government’s “motive” theory was “indistinguishable from predisposition.” Id. at 178. The evidence “invited [the jury] to infer from Harrison’s remarks to [the other girls] that he had a sexual interest in teenage girls, and ... that he acted in conformity with that bad character trait by committing the charged offenses against [the victim].” Id. at 180. We noted that
[t]he more common or generalized the motive evidence, the more it verges upon inadmissibility as mere propensity evidence. If, for instance, in a sexual assault prosecution, evidence of prior bad acts against other victims is introduced to show the defendant’s desire to engage in heterosexual sex, the motive is indistinguishable from predisposition — for such evidence to be relevant, the jurors must infer the defendant’s general sexual desire from the prior bad acts, and then infer that he acted in conformity with that desire and committed the charged sexual offense. As a rule, it is improper to offer prior instances of the same offense [against persons other than the victim in the instant case] to show motive, since such acts are relevant only by relying on the improper *1261inference that the defendant has a propensity to engage in that conduct.
Id. at 178 (alterations in original) (quotation marks omitted); see also Hill v. United States, 600 A.2d 58, 62 (D.C.1991) (“The key to admissibility under the motive exception ... is the fact that the defendant’s prior criminal conduct was directed toward the same victim.”).
Here, Thomas’s conduct in the Arlington, Virginia incident on December 31, 2008, some four months after the last incident Thomas was charged with in Georgetown, was not directed towards any of the victims in the Georgetown incidents.10 While we have recognized that there is “no reason to artificially distinguish between those situations where the victim of the initial wrongful conduct and the ultimate crime are identical, and where the ultimate victim is a third party with a clear nexus to the initial misconduct,” Mitchell v. United States, 629 A.2d 10, 14 (D.C.1993), no such clear nexus exists in the present case. In fact, the only “nexus” between the aggravated sexual assault conviction in Virginia and the Georgetown incidents the government is able to identify is that “[Thomas] was motivated to prey on intoxicated young white men in vulnerable positions.” This argument is indistinguishable from the motive proffered by the government in Harrison — that Harrison was motivated to seize an “opportunity” to assault “any girl” who happened to be so “unlucky.” Harrison, supra, 30 A.3d at 179 (quotation marks omitted). The government’s argument is therefore precisely the “common or generalized- [ ... ] motive evidence” that we held was “inadmissib[e] as mere propensity evidence” in Harrison. Id. at 178 (quotation marks omitted). See also Thompson v. United States, 546 A.2d 414, 421 (D.C.1988) (“[w]here evidence of prior crimes can become probative with respect to intent only after an inference of predisposition has been drawn, the argument for admission is at its weakest, for the distinction between intent and predisposition then becomes ephemeral”). For those reasons we reject it.
Both of the government’s other arguments — that the other-crimes evidence was relevant to show Thomas’s identity “with regard to the sexual assault of [B.C.]” and that the evidence was relevant to show the absence of mistake and state of mind “with respect to” the charges “involving [M.C.]” — presume that the aggravated sexual assault incident in Virginia was so distinctively similar to those two charged offenses that a jury could reasonably infer that Thomas committed both. We conclude that it was not.
For other-crimes evidence to be admitted under the identity exception, it must “show[ ] that the defendant has committed crimes so nearly identical in method that it is likely that the present offense has been committed by him.” Bridges v. United States, 381 A.2d 1073, 1075 (D.C.1977). “To be probative of intent ... the prior criminal conduct usually must involve an offense similar in kind and reasonably close in time to the charge at trial.” Willcher v. United States, 408 A.2d 67, 76 *1262(D.C.1979). “[T]he fact that intent is in issue is not enough to let in evidence of similar acts, unless they are so connected with the offense charged in point of time and circumstances as to throw light upon the intent.” Boyer v. United States, 76 U.S.App. D.C. 397, 398, 132 F.2d 12, 13 (1942) (quotation marks omitted).
Here, while Thomas did make his state of mind an issue, by claiming that the incident with M.C. in August 2008 was a misunderstanding, the circumstances surrounding the aggravated sexual assault incident in Virginia and the charged crime did not entail “a case of peculiar coincidence, in which mere recurrence throws light on mental states.” Id. The government fails to identify any significant similarities between the aggravated sexual assault in Virginia and the alleged assault on M.C.11
Moreover, the only similarities the government has identified between the aggravated sexual assault in Virginia and the alleged sexual assault of B.C., which occurred seventeen months before the Virginia incident, are that both involved sexual assaults of young white male victims late at night. There was no burglary of the Arlington victim’s home, as was the case involving B.C., since the sexual assault upon the Arlington victim took place in Thomas’s home in Virginia. In contrast, the charge involving B.C. occurred in Georgetown, after a burglary of the victim’s home. These two incidents are qualitatively different. A crime committed against a guest in one’s own home carries almost a certain likelihood of identification of the perpetrator by the victim, and may suggest that the perpetrator believed the act to be consensual and therefore not criminal. On the other hand, a crime committed in the course of a burglary carries no greater risk of identification of the perpetrator than the commission of the burglary itself. Further, the B.C. incident in Georgetown occurred in August 2008, while the Arlington incident occurred much later, at the end of December 2008, in a very different location. For these reasons, we are satisfied that the “totality of the factual circumstances” does not provide “a sufficient basis for admission under the Drew doctrine.” Easton, supra, 533 A.2d at 907 (emphasis and quotation marks omitted). We, accordingly, hold the trial court erred by admitting the other-crimes evidence. For the reason set forth below, we also conclude that the error was not harmless.
We apply the harmless error test of Kotteakos v. United States, 328 U.S. 750, 764-65, 66 S.Ct. 1239, 1248, 90 L.Ed. 1557, 1566-67 (1946) to the erroneous admission of prior crimes evidence. See Veney v. United States, 936 A.2d 811 (amended on other grounds, 936 A.2d 809 (D.C.2007)). The error is harmless only if one can 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. at 1248, 90 L.Ed. at 1566-67.
We have recognized that
[ejvidence of prior wrongful behavior is always prejudicial to a defendant. It not only risks that the jury may infer guilt simply on the basis that the accused has committed wrongful acts, but it diverts the jury’s attention from the question of defendant’s responsibility for *1263the crime charged to the improper issue of his bad character.
Campbell v. United States, 450 A.2d 428, 431 (D.C.1982).
In the context of the identity exception, we have noted that “[o]nce the identity exception to the other crimes rule has been rejected, the points of potential similarity cut the other way. Every suggestion at trial that the two crimes were in some way similar increased the likelihood that the jury became confused or misused the evidence.” Tinsley v. United States, 368 A.2d 581, 586-87 (D.C.1976) (quotation marks omitted).
In this instance, the jury heard from the Arlington victim that he woke up in an unknown apartment “laying on the floor on [his] back ... [with his] pants ... down” and Thomas was “fondling” and “masturbat[ing]” him. The victim suggested that he was helpless to move or stop the assault. After the assault, articles of his clothing were missing, and he later discovered his underwear was on backwards. This evidence was highly prejudicial, and “points of potential similarity” between the aggravated sexual assault in Virginia and the Georgetown crimes significantly “increased the likelihood that the jury became confused or misused the evidence,” particularly in light of B.C.’s sexual assault allegations.12 Id. at 536-37. Moreover, similar to Harrison, “[t]he court’s instruction not to consider that evidence to conclude that [Thomas] had a bad character or a criminal personality was not curative.” Harrison, supra, 30 A.3d at 180. We therefore conclude the trial court’s error was not harmless.
III.
Although we are reversing the convictions for the reasons set forth above, we will address the other claims because they are likely to arise at any re-trial. Thomas argues that the trial court abused its discretion by precluding him from presenting evidence that the Georgetown Cuddler might have committed the crimes that he was charged with. Specifically, Thomas claims that the third-party perpetrator evidence was admissible because those assaults against other Georgetown University students were committed in a similar manner, in the same locale, during the same time of morning, and the assaults continued even after he was incarcerated or confined through electronic monitoring.
“Winfield evidence” is evidence that “tends to show that someone other than the defendant was the real culprit.” Newman, supra, 705 A.2d at 254 (citing Winfield, supra, 676 A.2d at 1). “We review a trial court’s determination on the admissibility of a third-party perpetrator defense for abuse of discretion, and that determination ‘will be upset on appeal only upon a showing of grave abuse.’ ” Melendez v. United States, 26 A.3d 234, 241 (D.C.2011) (quoting Gethers v. United States, 684 A.2d 1266, 1271 (D.C.1996); McCraney v. United States, 983 A.2d 1041, 1050 (D.C.2009)).
For Winfield evidence to be admissible,
there must be “proof of facts or circumstances which tend to indicate some reasonable possibility that a person other than the defendant committed the charged offense.” The “focus” of the *1264standard is not on the third party’s guilt or innocence, but on “the effect the evidence has upon the defendant’s culpability,” and in this regard it “need only tend to create a reasonable doubt that the defendant committed the offense.”
Bruce v. United States, 820 A.2d 540, 543 (D.C.2003) (quoting Winfield, supra, 676 A.2d at 4 (emphasis omitted)). “[F]or admissibility the crimes need not be identical if the totality of the circumstances demonstrates a reasonable probability that the same man attacked both complainants.” Newman, supra, 705 A.2d at 257. However, the trial court should exclude Winfield evidence if it “is too remote in time and place, completely unrelated or irrelevant to the offense charged, or too speculative with respect to the third party’s guilt.” Resper v. United States, 793 A.2d 450, 460 (D.C.2002) (quotation marks omitted). Moreover, the proponent must show a nexus between the proffered evidence and the crime charged, and, as we stated in Win-field, “the determination that evidence is relevant does not exhaust the trial judge’s responsibility in deciding whether to admit it. The judge must also balance the probative value of the evidence against the risk of prejudicial impact.” 676 A.2d at 5 (quotation marks omitted). The trial court thus has the “discretion to exclude marginally relevant evidence” that “will distract the jury from the issue in this case.” Id.
Here, the trial court did not abuse its discretion by precluding Thomas from presenting evidence that the Georgetown Cuddler might have committed the crimes that he was charged with. The only similarities that Thomas identifies are that all the crimes involved burglaries that occurred in and around Georgetown early in the morning and that the perpetrator left when he was discovered without engaging in violent behavior. Under the totality of the circumstances, those similarities are insufficient by themselves to establish the nexus required between the proffered evidence and the crimes at issue. As Thomas concedes, all but one of the proffered Cud-dler incidents involved female victims in and around Georgetown, and as the trial court correctly found, “the one incident that [Thomas] ... cite[d] to where a man is involved, the suspect in that instance entered the house, apparently saw the man, and then left.” The proffered incidents were therefore markedly different from the crimes Thomas was charged with, which targeted male victims in two adjoining townhomes and in another townhome a few blocks away.
Thomas asserts that the same perpetrator “could have massaged a woman’s shoulders or he could have sat down next to her on the bed.” We think that claim is entirely speculative because the overly broad similarities Thomas cites do not establish the nexus required between the proffered evidence and the crimes at issue. In other words, the “totality of the circumstances” does not demonstrate a reasonable probability that the same man attacked both the female complainants in Thomas’s proffer and the male complainants in the crimes Thomas was convicted of. This conclusion is consistent with the decisions of this court addressing this type of evidence.
For example, in Winfield, the defendant was convicted for armed first-degree murder of a woman who had been cooperating with law enforcement against her former partners in crime. The trial court rejected a proffer of evidence that a third party actually committed the murder. 676 A.2d at 3-4. The evidence proffered showed that (1) identified third parties had a motive to silence the victim to prevent her testimony in court proceedings involving charges of armed robbery and abduction; (2) one of those third parties had threatened to kill the victim to prevent her from *1265“snitching”; (3) those third parties had tried to kill the victim a month before her death, demonstrating that they could find her; (4) given the timing, the reason for the murder likely was the victim’s cooperation with the authorities; (5) one of the third parties had the opportunity to kill the victim; (6) the method used was one of the methods used in the previous attempt to kill the victim; and (7) the killer’s reference to “snitching” during the crime related to the third parties’ motive, not the defendant’s. Id. at 6. On appeal, we reversed the trial court’s ruling rejecting defendant’s proffer, holding that the trial court should have admitted the proffered evidence because the aggregate evidence would, if proven, demonstrate more than coincidence of motive and prior assaults on the victim by the third party. Id.
In contrast, the similarities between the proffered evidence here and the crimes Thomas was charged with would have little probative value. The trial court correctly noted that “these other incidents that are dissimilar in terms of the conduct in the victims, most particularly the victims in those other instances, have no probative value.” As a result, the prejudicial effect of “distracting] the jury in this case,” outweighed its probative benefit to Thomas. We are therefore satisfied that the trial court did not abuse its discretion by precluding Thomas from presenting the third-party perpetrator evidence.
IV.
Finally, Thomas argues “that the trial court abused its discretion in excluding expert testimony on two [aspects] of eyewitness identification: (1) that exposure to post-event information can contaminate and undermine the reliability of an eyewitness identification in ways that are not apparent to a witness; and (2) that a witness’s expression of confidence is a poor proxy for the accuracy of the identification.”
Whether to
admit expert testimony is committed to the discretion of the trial court; a ruling either admitting or excluding such evidence will not be disturbed unless manifestly erroneous — i.e., for abuse of discretion. When an evidentiary question is committed to the discretion of the trial judge, as here, our review on appeal is limited to whether the judge engaged in a proper exercise of discretion. As we have said on many occasions, the exercise of discretion entails, first, recognition that there is discretion to be exercised, and then, after consideration of the correct legal factors, their reasonable application to the facts of the case before the court.
Benn v. United States, 978 A.2d 1257, 1273 (D.C.2009) (internal quotation marks omitted).
In Dyas, we identified three distinct criteria that trial judges must apply in considering whether to admit or exclude expert testimony regarding eyewitness identification:
(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.
Benn, 978 A.2d at 1269 (quoting Dyas v. United States, 376 A.2d 827, 832 (D.C. *12661977) . (internal quotation marks omitted and emphasis in original)).
[T]he court’s determination [to admit or exclude expert testimony] must be case-specific, based on the proffered expert testimony, and must consider: (1) the current state of generally-accepted scientific research; (2) whether it is within the common knowledge of lay jurors; and (3) whether the testimony would assist the jury, taking into account the relevance and probative value of the proposed scientific evidence to the eyewitness identification in the case.
Id. at 1278.
Here, the trial court concluded that Thomas’s expert’s testimony would not aid the trier of fact because post-event contamination and the correlation between confidence and accuracy, among other things, were not beyond the ken of the ordinary juror. In Benn, we recognized that “[Research reveals ... that the correlation between a witness’s expression of certainty in an identification and its accuracy is, at a minimum, greatly overstated, and perhaps unwarranted,” id. at 1268, and the “correlation between confidence and accuracy of an identification ... are counterintuitive.” Id. at 1277; see also Hager v. United States, 856 A.2d 1143, 1148, amended on other grounds, 861 A.2d 601 (D.C.2004) (“the correlation between witness confidence and accuracy ... may well be beyond the ken of the average layperson”). The Kassin study, which was attached to Thomas’s motion for reconsideration, showed that the impact of post-event information was common sense to 17% of jurors and the low correlation between accuracy and confidence was known to only 5% of jurors. Kassin, et al., supra at 412. The government implicitly concedes that the trial court’s ruling is inconsistent with this court’s opinion in Benn, and with the Kassin study, by failing to defend the trial court’s ruling on that ground.13 See McCrimmon, supra, 853 A.2d at 159 n. 8 (arguments not made in briefs on appeal are abandoned). Instead, the government argues that the trial court could have exercised its discretion to exclude the eyewitness identification testimony because the identifications were corroborated by other evidence.14 Because that was not the basis for the trial court’s ruling, we cannot affirm the trial court with respect to the admissibility of the expert’s testimony on that ground. We cannot sustain a ruling on a ground that the trial court did not rely on, unless there is only one way the trial court could have ruled as a matter of law.15 Wright v. United States, 508 A.2d 915, 919-20 (D.C.1986).16
*1267Ordinarily in these circumstances, we would remand the case and direct the trial court to reconsider its ruling in light of what we said in Berm and the Kassin study. The trial court could then address the issue of whether there was a basis for disallowing the testimony because the identifications were corroborated by other evidence. However, since we are reversing on the Drew issue, we assume that at any re-trial, the trial judge will consider this issue in light of the framework we have outlined here.
V.
Accordingly, we reverse Thomas’s convictions and remand the case for a new trial.
So ordered.
. Thomas was also convicted of charges related to tampering with a detection device. No challenge has been presented with regard to those charges.
. Winfield v. United States, 676 A.2d 1 (D.C. 1996) (en banc).
. The day after Thomas entered his bedroom, C.T. gave his friend J.K. a description of Thomas, and C.T. later told J.K. that he had seen the same man driving a tan or champagne-colored Lexus SUV. Still later, J.K. saw a man matching Thomas’s description, driving a tan or champagne-colored Lexus SUV, and he wrote down the SUV’s full license plate number. J.K. gave the license plate number to C.T., who, in turn, provided the full license plate number to Officer Pritchett.
. Thomas was also charged with entering the townhouse at 1207 33rd Street where R.L., a Georgetown University student, was staying and grazing R.L.'s thigh with his hand. How*1257ever, Thomas was acquitted of all charges involving R.L.
. B.C. also spoke with other complainants prior to making his photo identification.
. The Georgetown Cuddler evidence related to seven different incidents that occurred between September 2008 and March 2009.
. The jury was apparently not informed of Thomas' plea or the fact of his conviction. All evidence of this incident was introduced through the testimony of the victim.
. 118 U.S.App.D.C. at 15, 331 F.2d at 89.
. The government has not presented any arguments supporting the trial court’s admission of the other-crimes evidence on the basis of common scheme or plan. See McCrimmon v. United States, 853 A.2d 154, 159 n. 8 (D.C.2004) (arguments not made in briefs on appeal are abandoned).
. Arlington is a county of approximately twenty-five square miles in the state of Virginia which is located immediately across the Potomac River from the Georgetown section of the District of Columbia, directly connected by the Key Bridge. Fast Facts, Arlington County, http://www.arlingtonva.us/ departments/CPHD/planning/data_maps/ Census/CensusFacts.aspx (last visited Nov. 7, 2012). The record does not reveal, however, the distance between the Georgetown incidents and the Virginia incident, and the restaurant where the victim worked is apparently located in the Clarendon neighborhood, which is located approximately two miles from Georgetown. See D.C. Dept, of Public Works, Washington D.C. Transportation Map (1988 ed.).
. The only possible similarities we can discern on this record are that both incidents involved intoxicated young white male victims who were assaulted late at night. That is insufficient to meet the "fairly rigid standard of similarity.” Newman, supra, 705 A.2d at 255 (quotation marks omitted).
. While the prejudicial impact of propensity evidence can be minimized in some cases where the defendant is charged with multiple counts of similar criminal conduct, here Thomas questioned the evidence linking him to the scene of many of the charged crimes. Thus, in this case, evidence that Thomas had committed a single crime elsewhere had a large prejudicial impact.
. The government writes that "according to the Kassin report itself, only two of the appellant's proposed topics — the exposure to post-event information and the correlation between accuracy and confidence — were sufficiently reliable to support courtroom testimony." The government then argues that "in any event, even as to these two topics, the trial court did not abuse its discretion in precluding expert testimony in this case, where the eyewitness identifications were substantially corroborated.”
. In reaching our holding in this case, we are not expressing any opinion regarding the merits of this aspect of the government's position.
. The exception is inapplicable here, and the government does not claim otherwise.
. We cannot sustain a ruling that should have been discretionary, but was not, even though discretion, properly exercised, might have led to the same result. The essence of [discretionary] decision-making is the trial court's judgment in exercising that discretion, for a discretionary decision is based not only on hard facts but also— and often more importantly — on perceptions of demeanor or the pace of the trial, and, ultimately, of the probable impact of counsel and witnesses on the jury. The parties, therefore, are entitled to have the trial judge exercise that discretion, unfet*1267tered by erroneous legal thinking; they need not settle for the substituted judgment of an appellate court that would sustain the ruling on a plausible, alternative ground without benefit of all the data, derived from perceptions at trial, that inherently go into a discretionary ruling.
Wright, supra, 508 A.2d at 919-20 (D.C. 1986) (internal quotation marks and citations omitted).