United States v. Mejia

545 F.3d 179, 77 Fed. R. Serv. 1028, 2008 U.S. App. LEXIS 20992, 2008 WL 4459289

Opinion

HALL, Circuit Judge:

Appeal from judgment of conviction entered in the United States District Court for the Eastern District of New York (Wexler, J.) for conspiracy to commit assaults with a dangerous weapon in aid of racketeering activity, 18 U.S.C. § 1959(a)(6), three counts of assault with a dangerous weapon in aid of racketeering activity, id. § 1959(a)(3), and three counts of discharge of a firearm during a crime of violence, id. § 924(c)(1). We vacate the judgment after finding that the admission of expert witness Hector Alicea’s testimony violated the Federal Rules of Evidence and the Sixth Amendment Confrontation Clause, and that the error was not harmless. We remand for retrial.

BACKGROUND

I. The Drive-By Shootings

On June 18, 2003, Ledwin Castro and David Vasquez (collectively, “Appellants”), along with several others, participated in two drive-by shootings on Long Island, New York. At the time, Appellants were members of the MS-13 gang. 1 MS-13 is a nationwide criminal gang organized into local subunits known as “cliques.” At the time, to become a full member of MS-13, an individual was required to “make his quota,” which meant to engage in acts of violence against members of rival gangs, such as the SWP and the Bloods. MS-13 had local cliques on Long Island, and Castro was the leader of the Freeport clique (the “Freeport Locos Salvatruchas,” or “FLS”).

On the day of the shootings, Vasquez, Castro, Ralph Admettre (a member of MS-13), and Nieves Argueta, a new initiate into MS-13, met at the apartment of Bonerje Menjivar (also a member of MS-13). There, the group discussed their plan to shoot members of rival gangs. They had been preparing to carry out the shootings for quite some time. A few weeks earlier, Admettre, acting at Castro’s direction, had stolen the van that they would use. Earlier that day, Vasquez had informed Admettre that he — Vasquez—had procured a handgun belonging to the Free-port clique. While the group was at Men-jivar’s apartment, Menjivar gave Castro ammunition for the handgun.

Admettre, Castro, Vasquez, and Argueta put the plan into action that night. At about 9:40 p.m., Admettre drove the others to a laundromat in Hempstead, New York. After Vasquez and Castro reconnoitered the scene, attempting to determine whether anyone in the parking lot was a member of SWP, a rival gang, Admettre left the laundromat parking lot and stationed the van in a gas station parking lot across the street. Once the van was in position, Vasquez fired at the crowd in the laundromat parking lot from inside the van. Two individuals were hit. Ricardo Ramirez, age fifteen, was hit by three shots in the chest, arm, and leg. Douglas Sorto, age sixteen, was hit once in the leg. Both victims survived the shooting. After Vasquez fired these shots, Admettre drove away. Castro then called Menjivar and informed him that more ammunition would be needed. Admettre drove everyone back to Menjivar’s apartment, and Menjivar gave Vasquez additional ammunition.

Shortly thereafter, at approximately 10:20 p.m., Admettre, Castro, Vasquez, and Argueta traveled to a delicatessen parking lot in Freeport, New York. Upon arriving, they saw a group of young black men who they believed were members of the Bloods, a rival gang. Vasquez handed the handgun to Argueta, who proceeded to shoot one of the young men, Carlton Alexander, seven times in the back. Despite being hit by multiple shots, Alexander survived. After the shooting, the four men immediately abandoned the van. About one month later, local law enforcement arrested Castro, Vasquez, Admettre, and Argueta.

II. Indictment and Trial

In February 2004, a federal grand jury indicted Vasquez, Castro, and twelve others for various offenses stemming from a series of violent incidents on Long Island between August 2000 and September 13, 2003. A superseding indictment (“the Indictment”) was returned on June 23, 2005. The Indictment described MS-13, or “La Mara Salvatrucha,” as a gang that originated in El Salvador but had members throughout the United States. It accused all of the defendants of being members of MS-13, and it alleged that MS-13 members “engaged in criminal activity” in order to increase their position within the organization. According to the Indictment, MS-13 constituted an “enterprise” under 18 U.S.C. § 1959(b)(2) because it was an ongoing organization the activities of which affected interstate commerce. The Indictment furthermore stated that MS-13 engaged in two forms of racketeering activity under 18 U.S.C. § 1961(1): (1) acts and threats involving murder as defined by New York State law, and (2) narcoties trafficking as defined by federal law.

The Indictment charged both Appellants with ten counts. Count One charged them with conspiracy to commit assaults with a dangerous weapon in order to maintain and increase their position within the MS-13 racketeering enterprise, in violation of 18 U.S.C. § 1959(a)(6). Counts Six, Seven, and Eight charged Appellants with assaulting Ramirez, Sorto, and Alexander, respectively, with a dangerous weapon in order to maintain or increase their positions in the MS-13 racketeering enterprise, in violation of 18 U.S.C. § 1959(a)(3). Counts Twelve, Thirteen, and Fourteen accused Appellants of discharging a firearm during a crime of violence, in violation of 18 U.S.C. § 924(c)(l)(A)(iii). The Indictment identified the relevant crimes of violence as the assaults charged in Counts Six, Seven, and Eight. Finally, Counts Seventeen, Eighteen, and Nineteen charged Appellants with using an explosive to commit a felony — the felonies being the assaults charged in Counts Six, Seven, and Eight — in violation of 18 U.S.C. § 844(h)(1). 2

The district court severed the charges against Appellants from the charges against some of their co-defendants. Appellants’ remaining co-defendants pleaded guilty to assault, and Admettre pleaded guilty to the conspiracy charge and to using a firearm during a crime of violence. Appellants proceeded to trial before the district court. During the course of the trial, which took place between July 19 and July 26, 2005, the Government called Hector Alicea, an officer with the New York State Police, as an expert witness. The Government also called the three shooting victims and co-defendants Admettre and Menjivar. In addition, the Government introduced into evidence telephone records, the firearm used in the shootings, ballistics records, and Appellants’ post-arrest confessions.

Admettre testified to his membership in MS-13 and about the gang’s structure and operations. He stated that MS-13 was in “an all out war with rival gangs” — a war that included shootings, stabbings, fighting, and murder — and that MS-13 had a policy to murder the members of rival gangs. Admettre went on to identify Appellants as members of MS-13. He also described an uncharged shooting involving both Appellants that took place in February 2003:

David Vasquez identified [an] SWP member and opened fire. And Castro jumped out along with Yasquez to chase the SWF member down the block and both of them fired.

Later, Menjivar also testified about his membership in MS-13 and about the gang’s structure and operations. Menjivar further testified that the Freeport clique sent money to El Salvador to help members who had been deported from the United States. Alicea testified about MS-13’s history and structure, and he also explained MS-13’s activities on Long Island. Because the nature of Alicea’s testimony is an important and disputed issue on appeal, the subjects about which he testified are discussed further in our analysis of Appellants’ challenge to the admission of his testimony.

III. The Jury Verdict and Sentencing

On July 26, 2005, the jury found Appellants guilty on all ten counts. In a special verdict, the jury further found that MS-13 was an enterprise that affected interstate commerce; that MS-13 engaged in acts and threats of murder; that Appellants were members of the MS-13 enterprise; and that Appellants had participated in the conspiracy to assault and in the charged assaults in order to maintain or increase their positions within MS-13. The jury failed to find, however, that MS-13 engaged in the racketeering activity of narcotics trafficking.

One week later, Appellants asked the district court to set aside the verdict and enter a judgment of not guilty pursuant to Federal Rule of Criminal Procedure 29, or in the alternative to vacate the judgment and order a new trial pursuant to Federal Rule of Criminal Procedure 33. They claimed multiple errors. First, they claimed that the jury’s failure to find that MS-13 engaged in narcotics trafficking was fatal to the verdict because the Indictment had not pleaded the two racketeering activities (murder and narcotics trafficking) in the alternative. Second, they argued that the proof of MS-13’s involvement in murder was insufficient. Third, they asserted that the Government had failed to prove that the MS-13 enterprise had an existence separate from the charged offenses. The district court docket does not indicate when or how the district court denied the motion, but that it did so is clear from the progression of the case to sentencing.

On December 5, 2005, the district court sentenced Vasquez principally to a total of 63 years’ imprisonment. The sentence consisted of 3 years’ concurrent imprisonment for the conspiracy count (Count One) and each of the three assault charges (Counts Six, Seven, and Eight), as well as a consecutive 10-year term for the first firearm count (Count Twelve) and two additional consecutive 25-year terms for each of the other two firearm counts (Counts Thirteen and Fourteen). The district court imposed longer sentences for Counts Thirteen and Fourteen because it considered them to be second or subsequent firearm offenses. The court, acting sua sponte, dismissed the three explosive counts (Counts Seventeen, Eighteen, and Nineteen), ostensibly on the theory that those counts were in the nature of lesser-included offenses of the firearm counts. Judgment entered on December 13, 2005.

On January 6, 2006, the district court sentenced Castro principally to 60 years plus 1 day of imprisonment. The 1 day consisted of concurrent 1-day sentences for the conspiracy count (Count One) and the three assault charges (Counts Six, Seven, and Eight). As had been the case for Vasquez, the total of 60 years’ imprisonment was imposed for the three firearm offenses. The district court also dismissed the three explosive counts, for the same reasons it did so in Vasquez’s case.

IV. Arguments on Appeal

On appeal, Appellants challenge their convictions and sentences on multiple grounds. With one exception, Appellants bring these challenges jointly. They devote the bulk of their argument to their claim that the district court erred in allowing the Government to call Alicea as an expert witness and to their further claim that Alicea’s testimony violated the Federal Rules of Evidence and Crawford v. Washington, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004). Beyond their challenge to Alicea’s testimony, Appellants argue that the district court erred in admitting evidence of narcotics trafficking and an uncharged shooting, that the Government failed to prove that MS-13 was a racketeering enterprise, and that the three firearm counts were duplicative. They also object to their sentences on the second and third firearm offenses (Counts Thirteen and Fourteen), which the district court considered to be second or subsequent offenses, because the Indictment failed to charge those offenses as second or subsequent offenses. Castro raises one additional challenge on his own, arguing that the two assault charges stemming from the Hempstead shootings were multi-plieitous because the Government failed to introduce evidence that Vasquez intended to shoot more than one victim. Because the parties focused both at trial and on appeal on questions regarding Alicea’s testimony, we begin with that issue.

DISCUSSION

I. Hector Alicea’s Testimony as an Expert Witness

A. The Substance of Alicea’s Testimony

The Government called Hector Alicea, an investigator with the New York State Police, to testify regarding MS-13’s “enterprise structure and the derivation, background and migration of the MS-13 organization, its history and conflicts,” as well as MS-13’s “hierarchy, cliques, methods and activities, modes of communication and slang.” Alicea had been an officer of the New York State Police for eighteen years, and he had been an investigator since 1992. In June 2000, five years before the trial, Alicea had been assigned to the FBI Long Island Gang Task Force. He was also the Chair of the Intelligence Committee of the East Coast Gang Investigators Association.

Prior to trial, Appellants objected to the Government’s stated plan to call Alicea on the ground that Alicea would rely on “impermissible hearsay” to reach his conclusions. The accompanying memorandum of law cited to this Court’s opinion in United States v. Dukagjini, 326 F.3d 45 (2d Cir. 2003). The parties argued the motion before the district court, and the district court reserved its decision. Defense counsel continued to press the issue at trial, and Vasquez’s attorney was permitted to conduct a voir dire examination of Alicea prior to direct examination by the Government. In response to defense counsel’s questioning, Alicea stated that he had participated in somewhere between fifteen and fifty custodial interrogations of MS-13 members. When asked whether he could distinguish between information he had learned during custodial interrogations and information he had learned elsewhere, Alicea responded that his knowledge was based on “a combination of both.” At the conclusion of the voir dire, defense counsel again argued that Alicea was not qualified as an expert and that his testimony would introduce testimonial evidence in violation of Crawford. See 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177. The district court then denied the motion.

Much of Alicea’s testimony concerned MS-13’s background. He testified about MS-13’s history, its presence on Long Island, and its national and international presence; about the gang’s colors, hand signs, graffiti use, naming practices, and tattoos; and about its local subunit structure, leadership structure, division of responsibilities, and membership rules. In addition, Alicea testified to more specific details about MS-13’s operations. He stated that when MS-13 members fled from prosecution or needed to travel for “gang business reasons,” such as “to transport narcotics,” “to transport weapons,” or “to commit crimes in other areas,” they traveled “on a Greyhound bus” or by car. According to Alicea, MS-13 members from Virginia, California, and El Salvador had attended organizational meetings in New York State, and MS-13 leaders throughout the nation communicated with each other by telephone. He testified that MS-13 treasury money “[was] used to buy guns,” to help members in prison or in other states, or to buy narcotics. Significantly, Alicea also asserted that MS-13 needed guns “to do what MS 13 does, which is, you know, shoot at rival gang members, and sometimes in the process, obviously, some people get hit.”

With respect to MS-13’s activities on Long Island, Alicea testified that since he had joined the Task Force in June 2000, the Task Force had seized “[p]robably between 15 and 25” firearms from MS-13 members. He further testified that Task Force members had seized ammunition, manufactured outside of New York State, from MS-13 members on Long Island. Moving on to MS-13’s narcotics-related operations, Alicea told the jury that MS-13 members on Long Island had been arrested for dealing narcotics, primarily cocaine, and that the gang also occasionally dealt marijuana. Alicea also stated that MS-13 “tax[ed]” non-gang drug dealers who wished to deal drugs in bars controlled by MS-13. Most importantly, Alicea attested that MS-13 had committed “between 18 and 22, 23” murders on Long Island between June 2000 and the trial.

On cross-examination, defense counsel probed the sources of Alicea’s information. Because of the importance of Alicea’s answers, we quote from his testimony at length:

Q.....I thought you mentioned FLS ... funded itself at the beginning from the sale of marijuana?

A. No. I was referring to the MS-13 gang as a whole.

Q. Is it fair to say that somebody told you that?

A. I had read that from some of the articles that I had researched.

Q. Newspaper articles?

A. Reports from other law enforcement personnel.

Q. You also told us that MS members ... put a tax on narcotics sales in certain bars; is that correct?

A. I was told that by a gang member, yes. 3

Q. And I believe you told us that some of those [membership] dues were used to purchase narcotics?

A. That’s correct.

Q. Is it fair to say that that was told to you also by somebody who was in custody?

A. In custody and some that were not. Q. Well, can you tell me ... how many people in custody told you in substance that monies collected were used for narcotics?

A. Probably like a dozen.

A.....I said I am aware that there has been contact between California and New York.

Q. And how did you become aware of it?

A. Listening to recordings.

Q. And you stated that you got information with regard to MS-13s involved with Mexican drug cartels; is that correct?

A. Yes.

Q. And where did you get that information from?

A. From research on the Internet.

Q. Do you know the source of that information on the Internet?

A. Not off the top of my head. But I did retrieve that.

Q. Is it law enforcement or reporters?

A. I think it is a combination of a reporter doing a story and having a conversation with law enforcement.

Q. ... [WJith regard to the involvement of MS-1S [unth] the Mexican cartels or Colombian cartels, you never interviewed anybody who told you that, it was strictly from the Internet; is that correct?

A. That is correct.

B. The Emergence of the Officer Expert

Under Federal Rule of Evidence 702, in those situations where “scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue,” testimony by “a witness qualified as an expert by knowledge, skill, experience, training, or education” is permissible so long as “(1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case.” Fed.R.Evid. 702. The broad phrasing of the description “scientific, technical, or other specialized knowl edge” brings within the scope of the Rule both “experts in the strict sense of the word,” such as scientists, and “the large group sometimes called ‘skilled’ witnesses, such as bankers or landowners testifying to land values.” Id. advisory committee’s note. On the question of when expert testimony is appropriate, the Advisory Committee Notes refer to the traditional common law rule that expert testimony is called for when the “untrained layman” would be unable intelligently to determine “the particular issue” in the absence of guidance from an expert. Id. advisory committee’s note (quoting Mason Ladd, Expert Testimony, 5 Vand. L.Rev. 414, 418 (1952)) (internal quotation marks omitted).

In the 1980s, a new type of “skilled witness” began emerging: the law enforcement officer. In criminal cases, the Government began calling law enforcement officers to testify as experts on what we referred to as “the nature and structure of organized crime families.” United States v. Daly, 842 F.2d 1380, 1388 (2d Cir.1988). This Court first reviewed a challenge to the use of such an expert in United States v. Ardito, 782 F.2d 358 (2d Cir.1986). The Government had called an FBI agent to testify as an expert about terms such as “captain,” “capo,” “regime,” and “crew.” Id. at 363. We upheld the admission of that expert testimony because it “aided the jury in its understanding of’ recorded conversations between the two defendants. Id. Furthermore, we noted, the district court had reminded the jury that the defendants there had not been charged with any conduct relating to organized crime. Id.

One year later, we upheld the admission of expert testimony by a law enforcement officer on the related matter of the meaning of messages written in code. United States v. Levasseur, 816 F.2d 37, 45 (2d Cir.1987). Upholding such testimony was consistent with pre-Ardito cases where we and other Circuits had allowed law enforcement officers to testify as experts about the meaning of jargon relating to narcotics trafficking. E.g., United States v. Borrone-Iglar, 468 F.2d 419, 421 (2d Cir.1972) (upholding a law enforcement officer’s testimony “concerning the narcotics vernacular used in [recorded] telephone conversations”); see also United States v. Theodoropoulos, 866 F.2d 587, 590-91 (3d Cir.1989) (describing testimony “concerning the meaning of ... coded conversations” as “the paradigm situation for expert testimony under Rule 702”) (overruled on other grounds).

In subsequent years, we have encountered novel uses of these “officer experts” and approved of their testifying on a broader range of issues. For example, in United States v. Daly, 842 F.2d 1380 (2d Cir.1988), where the defendants were charged with “various crimes arising out of activities of the Gambino crime family,” we upheld the expert testimony of an FBI agent who “identified the five organized crime families that operate in the New York area” and “described their requirements for membership, their rules of conduct and code of silence, and the meaning of certain jargon.” Id. at 1383, 1388. After the Government had played surveillance tapes for the jury, the agent interpreted the jargon the speakers had used. Id. at 1384. This Court upheld the district court’s decision to admit the agent’s testimony, finding that the agent had testified about “much that was outside the expecta-ble realm of knowledge of the average juror.” Id. at 1388. The district court’s judgment that the agent’s testimony would be helpful to the juror “was not unreasonable.” Id. Finally, the agent had not testified about the defendants or any of the charged offenses. Id. The only offense element to which the agent had testified “was the existence of a RICO enterprise, as he gave his understanding of the existence of organized crime and the Gambino family.” Id.

Since Daly, we have repeatedly upheld the admission of similar testimony. See, e.g., United States v. Locascio, 6 F.3d 924, 936 (2d Cir.1993) (upholding an FBI agent’s expert testimony about the internal operating rules of organized crime families, the meaning of recorded conversations, and the identification of members of the Gambino crime family); United States v. Feliciano, 223 F.3d 102, 109 (2d Cir.2000) (upholding an FBI agent’s expert testimony about “the structure, leadership, practices, terminology, and operations of [a street gang, Los Solidos]”); United States v. Matera, 489 F.3d 115, 121 (2d Cir.2007) (upholding the admission of an officer’s expert testimony “about the composition and structure of New York organized crime families” and observing that the district court had limited the expert’s testimony to general information rather than information about the defendants themselves). Our decision to permit such expert testimony reflects our understanding that, just as an anthropologist might be equipped by education and fieldwork to testify to the cultural mores of a particular social group, see Dang Vang v. Toyed, 944 F.2d 476, 481-82 (9th Cir.1991) (upholding the district court’s admission of expert testimony on Hmong culture), law enforcement officers may be equipped by experience and training to speak to the operation, symbols, jargon, and internal structure of criminal organizations. Officers interact with members of the organization, study its operations, and exchange information with other officers. As a result, they are able to break through the group’s antipathy toward outsiders and gain valuable knowledge about its parochial practices and insular lexicon. Allowing law enforcement officers to act as experts in cases involving these oft-impenetrable criminal organizations thus responds to the same concerns that animated the enactment of the criminal laws that such organizations (and them members) are typically charged with violating, such as the Racketeer Influenced and Corrupt Organizations Act, 18 U.S.C. §§ 1961-68, and the more recent Violent Crimes in Aid of Racketeering Act, id. § 1959. See Organized Crime Control Act of 1970, Pub.L. 91-452 pmbl., 84 Stat. 922, 923 (1970) (“[Organized crime continues to grow because of defects in the evidence-gathering process of the law inhibiting the development of the legally admissible evidence necessary to bring criminal ... sanctions ... to bear on the unlawful activities of those engaged in organized crime....”).

Yet despite the utility of, and need for, expertise of this sort, its use must be limited to those issues where sociological knowledge is appropriate. An increasingly thinning line separates the legitimate use of an officer expert to translate esoteric terminology or to explicate an organization’s hierarchical structure from the illegitimate and impermissible substitution of expert opinion for factual evidence. If the officer expert strays beyond the bounds of appropriately “expert” matters, that officer becomes, rather than a sociologist describing the inner workings of a closed community, a chronicler of the recent past whose pronouncements on elements of the charged offense serve as shortcuts to proving guilt. As the officer’s purported expertise narrows from “organized crime” to “this particular gang,” from the meaning of “capo” to the criminality of the defendant, the officer’s testimony becomes more central to the case, more corroborative of the fact witnesses, and thus more like a summary of the facts than an aide in understanding them. • The officer expert transforms into the hub of the case, displacing the jury by connecting and combining all other testimony and physical evidence into a coherent, discernible, internally consistent picture of the defendant’s guilt.

In such instances, it is a little too convenient that the Government has found an individual who is expert on precisely those facts that the Government must prove to secure a guilty verdict — even more so when that expert happens to be one of the Government’s own investigators. Any effective law enforcement agency will necessarily develop expertise on the criminal organizations it investigates, but the primary value of that expertise is in facilitating the agency’s gathering of evidence, identification of targets for prosecution, and proving guilt at the subsequent trial. When the Government skips the intermediate steps and proceeds directly from internal expertise to trial, and when those officer experts come to court and simply disgorge their factual knowledge to the jury, the experts are no longer aiding the jury in its factfinding; they are instructing the jury on the existence of the facts needed to satisfy the elements of the charged offense. See United States v. Nersesian, 824 F.2d 1294, 1308 (2d Cir.1987) (“In the past, we have upheld the admission of expert testimony to explain the use of narcotics codes and jargon.... We acknowledge some degree of discomfiture [when] this practice is employed, since, uncontrolled, such use of expert testimony may have the effect of providing the government with an additional summation by having the expert interpret the evidence.”). It is as though the law enforcement agency in question is a standing master for the criminal court, and the officer expert its representative charged with reporting that master’s findings of fact. Not only are masters a creature of civil rather than criminal courts, see Amalia D. Kessler, Our Inquisitorial Tradition: Equity Procedure, Due Process, and the Search for an Alternative to the Adversarial, 90 Cornell L.Rev. 1181, 1200, 1204 (2005) (describing the advent of masters in fifteenth-century English courts of equity), that sort of usurpation of the jury’s role is unacceptable even in the civil context, see James Wm. Moore, 9 Moore’s Federal Practice § 53.13[1], at 53-78 (3d ed. 2005) (“The 2003 amendments [to the Federal Rules of Civil Procedure] abolish the authority of trial courts to appoint trial masters respecting matters to be decided by a jury unless a statute provides otherwise.”). The Government cannot satisfy its burden of proof by taking the easy route of calling an “expert” whose expertise happens to be the defendant. Our occasional use of abstract language to describe the subjects of permissible officer expert testimony, e.g., Locascio, 6 F.3d at 936 (“We have ... previously upheld the use of expert testimony to help explain the operation, structure, membership, and terminology of organized crime families.”); United States v. Lombardozzi, 491 F.3d 61, 78 (2d Cir.2007) (“This Court has also permitted expert testimony regarding the organization and structure of organized crime families in [RICO] prosecutions .... ”), cannot be read to suggest otherwise.

This Court has not been blind to these risks. More than fifteen years ago, we observed that although “the operations of narcotics dealers are a proper subject for expert testimony under Fed.R.Evid. 702, we have carefully circumscribed the use of such testimony to occasions where the subject matter of the testimony is beyond the ken of the average juror.” United States v. Castillo, 924 F.2d 1227, 1232 (2d Cir. 1991) (citations omitted); see also United States v. Tapia-Ortiz, 23 F.3d 738, 740 (2d Cir.1994) (cautioning that expert testimony relating to “the operations of narcotics dealers ... should normally be used only for subjects that have esoteric aspects rea sonably perceived as beyond the ken of the jury” (internal quotation marks omitted)). Two years later, in Locascio, after approving of an FBI agent’s expert testimony on the structure of the Gambino family, we nonetheless “remind[ed] the district courts ... that they are not required to admit such testimony, and when they do the testimony should be carefully circumscribed to ensure that the expert does not usurp either the role of the judge in instructing on the law, or the role of the jury in applying the law to the facts before it.” Locascio, 6 F.3d at 939. In most cases, of course, no reminder was needed, often because the district court had taken affirmative steps on its own to prevent such usurpation. See, e.g., Ardito, 782 F.2d at 363 (noting that the district court had “specifically cautioned the jury as to the limited purpose of the agent’s testimony”); Daly, 842 F.2d at 1389 (“[T]he final [jury] instructions made clear that it was the jury’s province to determine whether or not the individuals named in the indictment functioned as an ‘enterprise’.... ”); Matera, 489 F.3d at 121 (“Immediately after [the officer expert’s] testimony, the district court gave a limiting instruction.... ”).

We more recently cautioned the Government of our concern about these risks in United States v. Dukagjini, 326 F.3d 45 (2d Cir.2003), and Lombardozzi, 491 F.3d 61. In Dukagjini, the Government called the case agent, a DEA officer, as an expert witness for the purpose of interpreting recorded conversations. 326 F.3d at 49-50. The district court allowed the agent to testify, but it “cautioned the prosecutor to limit [the agent’s] testimony to ‘words of the trade, jargon,’ and general practices of drug dealers.” Id. at 50. During his testimony, the agent interpreted various terms, such as “dry” and “cooked.” Id. The agent also addressed specific exchanges in recorded conversations and explained their meaning to the jury. Id. In doing so, the agent relied on both his experience and his knowledge of the case. Id.

After reviewing the agent’s testimony, we concluded that the agent had “stray[ed] from his proper expert function” by “act[ing] at times as a summary prosecution witness.” Id. at 55. The Government’s decision to call the case agent as an expert witness, we observed, had “increase[d] the likelihood that inadmissible and prejudicial testimony [would] be proffered.” Id. at 53. The officer expert’s status, we suggested, was likely to give his factual testimony an “unmerited credibility” before the jury. Id.; see also United States v. Alvarez, 837 F.2d 1024, 1030 (11th Cir.1988) (“When the expert is a government law enforcement agent testifying on behalf of the prosecution about participation in prior and similar cases, the possibility that the jury will give undue weight to the expert’s testimony is greatly increased.”). The defense’s inability to meaningfully challenge the case agent’s expert opinions inadvertently reinforces the agent’s credibility on questions of fact. Dukagjini, 326 F.3d at 53-54. In addition, case agents testifying as experts are particularly vulnerable to making “sweeping conclusions” about the defendants’ activities. Id. at 54.

We have identified two distinct ways in which the officer expert might “stray from the scope of his expertise.” Id. at 55. The expert might, as did the agent in Dukagjini, “testif[y] about the meaning of conversations in general, beyond the interpretation of code words.” Id.-, see also United States v. Freeman, 488 F.3d 1217, 1227 (9th Cir.2007) (“The fact that [the officer expert] possessed specialized knowledge of the particular language of drug traffickers did not give him carte blanche to testify as to the meaning of other words in recorded telephone calls without regard to reliability or rele- vanee.”). Or, we noted, the expert might “interpret[ ] ambiguous slang terms” based on knowledge gained through involvement in the case, rather than by reference to the “fixed meaning” of those terms “either within the narcotics world or within this particular conspiracy.” Dukagjini, 326 F.3d at 55.

We went on to find that, because the officer expert had relied on hearsay and custodial interrogations when forming his opinions, his testimony that went outside the scope of his expertise violated the Federal Rules of Evidence and the Confrontation Clause of the Sixth Amendment. Id. at 58-59. When the agent “departed from the bounds of Rules 702 and 703” by “repeatedly deviat[ing] from his expertise on drug jargon,” he thereby “crossed [the] line” between “permissible and impermissible reliance on hearsay.” Id. at 58-59. We held that the agent’s testimony violated the rules governing expert witnesses, id. at 55, the hearsay rules, id. at 59, and the Confrontation Clause, id. We affirmed the convictions nonetheless because the Confrontation Clause violation had not been plain error, id. at 61, and the hearsay violation had been harmless, id. at 62.

Similar concerns arose in Lombardozzi, 491 F.3d 61. There, the defendant had been charged with loan sharking, and the Government called an investigator with the U.S. Attorney’s Office for the Southern District of New York “as an expert who testified as to, inter alia, the general structure of La Cosa Nostra in New York and Lombardozzi’s affiliation with organized crime.” Id. at 72. In addition to testifying about the structure of La Cosa Nostra, the officer told the jury that the defendant was “a soldier in the Gambino crime family.” Id. When questioned by defense counsel as to his basis for that opinion, the witness testified that his knowledge of the defendant’s position in the Gambino family “was based on conversations with cooperating witnesses and confidential informants.” Id. He added that “he personally observed Lombardozzi’s activities approximately two dozen times since 1985.” Id. Because the defendant had failed to raise a Confrontation Clause challenge to the officer expert’s testimony, we reviewed that testimony for plain error and found that it did not affect the defendant’s substantial rights. Id. We commented, however, that “the record indicate[d] that” the expert may have “communicated out-of-court testimonial statements of cooperating witnesses and confidential informants directly to the jury in the guise of an expert opinion.” Id.

It is in light of these concerns that we now turn to Appellants’ specific challenges to Alicea’s testimony. In doing so, we review the district court’s admission of expert testimony for abuse of discretion, and we will not find error unless the district court’s ruling was “manifestly erroneous.” Dukagjini, 326 F.3d at 52 (quoting Locascio, 6 F.3d at 936).

C. Alicea’s Qualifications as an Expert

Appellants’ first challenge to Alicea’s testimony is a claim that he was unqualified to testify as an expert. They argue that because much of Alicea’s background is in the area of narcotics rather than gangs, he is not qualified to testify about the operations and structure of MS-13.

Under Rule 702, the district court may admit expert testimony if the witness is “qualified as an expert by knowledge, skill, experience, training, or education.” Fed.R.Evid. 702. At trial, Alicea testified that he had been an officer with the New York State Police for eighteen years and that he had been an investigator since 1992. He had been trained at the New York State Police academy, and he had received additional training in the field and through refresher courses. In addition, he had been a member of the FBI Gang Task Force on Long Island since 2000 and served as Chair of the Intelligence Committee of the East Coast Gang Investigators’ Association. Alicea also had extensive experience relating to MS-13 in particular. He testified that he had “listened to many conversations on tape,” performed surveillance in the course of investigations of MS-13, executed search warrants, debriefed MS-13 members, and trained other police departments on MS-13. Since joining the Task Force, he had arrested “between 50 and a hundred” and interviewed “[ojver a hundred” MS-13 members. He said that he had also “read a lot of documents related to MS, either on the internet or the media or from our instructors or other people from the conferences I have gone to,” and that he read “a web site” dealing with MS-13 on a daily basis.

Alicea’s qualifications are quite similar to those of experts whose qualifications we have upheld in the past. In Locascio, for example, the officer expert “had been an FBI agent for seventeen years, and for five years had been on the FBI’s Organized Crime Program, a squad that investigated only organized crime cases.” 6 F.3d at 937. Similarly, Alicea had been an investigator with the New York State Police for thirteen years and a member of the Task Force for five. Likewise, in Mat-era, the officer expert had “extensive experience investigating organized crime as a New York Police Department Detective and later as an Investigator for the United States Attorney’s Office.” 489 F.3d at 122. And in Feliciano, we described an officer expert as having “extensive experience” with a particular criminal gang based on the officer’s participation in a joint task force for approximately five years, “execution of federal and state search warrants at [gang] locations,” participation in electronic surveillance, and review of reports by other agents. 223 F.3d at 109. Ali-cea’s work experience, training, and involvement in investigations of MS-13 are at least as “extensive.” The district court did not err in finding that Alicea was qualified to testify as an expert.

The problem was not Alicea’s qualifications. It was the subjects about which he testified and the sources on which he relied. We address those concerns below.

D. The Appropriate Boundaries of Alicea’s Expertise Under Rule 702

Appellants argue that some of the matters about which Alicea testified were outside the scope of his expertise. Rule 702 requires that expert testimony concern “scientific, technical, or other specialized knowledge.” Fed.R.Evid. 702. Testimony is properly characterized as “expert” only if it concerns matters that the average juror is not capable of understanding on his or her own. See United States v. Amuso, 21 F.3d 1251, 1263 (2d Cir.1994) (“A district court may commit manifest error by admitting expert testimony where the evidence impermissibly mirrors the testimony offered by fact witnesses, or the subject matter of the expert’s testimony is not beyond the ken of the average juror.”); Locascio, 6 F.3d at 936 (applying the “untrained layman” standard articulated in the Advisory Committee Notes to Rule 702).

Much of Alicea’s testimony concerned material well within the grasp of the average juror. A few examples are particularly striking: Alicea’s testimony that the FBI gang task force had seized “[p]roba-bly between 15 and 25” firearms, as well as ammunition, from MS-13 members; his statement that MS-13 members on Long Island had been arrested for dealing nar- cotíes; and his statement that MS-13 had committed “between 18 and 22, 23” murders on Long Island between June 2000 and the trial. No expertise is required to understand any of these facts. Had the Government introduced lay witness testimony, arrest records, death certificates, and other competent evidence of these highly specific facts, the jury could have “intelligently” interpreted and understood it. For example, in United States v. Feliz, 467 F.3d 227 (2d Cir.2006), where the defendant was charged with RICO murder, “[i]n order to establish the manner and cause of death for each of [the defendant’s] victims in the charged homicides, the Government offered nine autopsy reports through the testimony of [a medical examiner],” id. at 229. The Government could have done the same here. Expert testimony might have been helpful in establishing the relationship between these facts and MS-13, but it was not helpful in establishing the facts themselves.

In addition to these stark examples, much of the remainder of Alicea’s testimony also addressed matters that the average juror could have understood had such factual evidence been introduced. Alicea’s testimony about the ways that MS-13 members traveled when fleeing from prosecution or when transporting contraband, his assertion that MS-13 members from Virginia, California, and El Salvador had attended organizational meetings in New York State, and his statement that MS-13 leaders communicated by telephone all fall into this category. So, too, did his testimony about the use of MS-13 treasury funds to buy firearms and narcotics, his statement that the gang occasionally dealt narcotics, and his statement that MS-13 taxed non-member drug dealers.

This testimony, which ranged from MS-13’s activities on Long Island to aspects of the gang’s operations more generally, went far beyond interpreting jargon or coded messages, Ardito, 782 F.2d at 363; Levasseur, 816 F.2d at 45, describing membership rules, Daly, 842 F.2d at 1388, or explaining organizational hierarchy, Locascio, 6 F.3d at 936. We find especially disturbing the portion of Alicea’s testimony that essentially summarized the results of the Task Force investigation on Long Island, and in particular Alicea’s testimony that MS-13 had committed between eighteen and twenty-three murders since 2000.

We recognize that expertise may have been necessary to connect specific murders to MS-13. An appropriate (admissible) example of such expertise would have been an expert’s explanation of how the graffiti near a body indicated that the murderer was a member of MS-13, or an expert’s testimony that the gang used a particular method to kill enemies and that as a result of his review of the autopsy reports (which would have been in evidence before the jury), he had concluded that MS-13 committed those murders. The acceptable use of expert testimony for this limited purpose, however, does not make it acceptable to substitute expert testimony for factual evidence of murder in the first instance (a fact, as we must remember, that is an element of the charged offense). If the Government is going to use the fact of murders as a way of proving that the gang engaged in a pattern of racketeering activity involving murder, that an individual was murdered remains a fact that must be proven by competent evidence. Only then does expertise of the sort proffered by Alicea become necessary to help the jury understand the particular evidence and to show a connection between MS-13 and the murder. The Government cannot take a shortcut around its obligation to prove murder beyond a reasonable doubt just by having an expert pronounce that unspecified deaths of eighteen to twenty-three persons have been homicides committed by members of MS-13. Alicea’s testimony in this regard went beyond those issues on which his “expert” testimony would have been helpful and appropriate.

In Feliciano, the one case where we have approved of testimony that even arguably approached the scope of Alicea’s testimony here, the officer was testifying as both a fact witness and an expert witness. 223 F.3d at 121. Like Appellants, the defendants in Feliciano had been charged with offenses under the Violent Crimes in Aid of Racketeering Act. Id. at 107. The racketeering enterprise of which those defendants were members was the Los Solidos gang. Id. At trial, the Government called the coordinator of a joint task force on gang activity in Hartford, Connecticut, as both an expert witness and a fact witness. Id. at 109-10. As an expert, the agent testified about “the structure, leadership, practices, terminology, and operations of Los Solidos.” Id. at 109. As a fact witness, he testified about his involvement in the task force investigation of Los Solidos. Id. at 109-10. On appeal, the defendants argued that the agent had testified to legal conclusions. Id. at 120-21. They did not challenge the agent’s testimony as outside the scope of his expertise. We upheld the district court’s admission of the testimony after finding that the agent had not testified to any legal conclusions. Id. at 121. Even as we affirmed the convictions, however, we expressed concern that “the line between [the agent’s] opinion and fact witness testimony [was] often hard to discern, and the ‘facts’ testified to are often stated very broadly and generally.” Id. Nonetheless, because the defense had had ample opportunity to cross-examine the agent as to the basis for those broadly stated facts, we found that the district court had not manifestly erred in admitting the testimony.

Feliciano thus reinforces our conclusion that it would be improper for the Government to rely on an officer expert’s testimony about matters outside the scope of any conceivable expertise and that the district court errs in allowing it to do so. There, the Government correctly realized that it could not call the task force coordinator to testify as an expert about the precise operations of the racketeering enterprise charged in the indictment, and it had the agent testify in dual roles: as a fact witness and as an expert witness. Alicea, in contrast, was proffered and testified in the ease before us only as an expert. Those parts of his testimony that involved purely factual matters, as well as those in which Alicea simply summarized the results of the Task Force investigation, fell far beyond the proper bounds of expert testimony. Alicea was acting as a de facto “case agent” in providing this summary information to the jury (the case being the ongoing investigation into MS-13’s activities on Long Island).

Testifying as he did, Alicea’s evidence runs afoul of our admonition in Dukagjini. When case agents testify as experts, they gain “unmerited credibility when testifying about factual matters from first-hand knowledge.” Dukagjini, 326 F.3d at 53. The testimony loses its expert character and the entire process transforms into “the grand jury practice, improper at trial, of a single agent simply summarizing an investigation by others that is not part of the record.” Id. at 54. Alicea’s factual testimony about matters that required no specialized knowledge clearly implicates these concerns, and the district court erred in allowing him to testify beyond the bounds for which expert testimony would have assisted the jury in understanding the evidence.

E. Alicea’s Reliance on Inadmissible Evidence Under Rule 703

At trial and on appeal, Appellants also claim that Alicea impermissibly relied on inadmissible hearsay in forming his conclusions.

Under Rule 703, experts can testify to opinions based on inadmissible evidence, including hearsay, if “experts in the field reasonably rely on such evidence in forming their opinions.” Locascio, 6 F.3d at 938; accord Fed.R.Evid. 703. Alicea unquestionably relied on hearsay evidence in forming his opinions. This hearsay evidence took the form of statements by MS-13 members given in interviews, both custodial and noncustodial, as well as statements made by other law enforcement officers, statements from intercepted telephone conversations among MS-13 members (which may or may not have been hearsay, depending on whether the conversations were in the course of and in furtherance of the charged conspiracies, see Fed.R.Evid. 801(d)(2)(E)), and printed and online materials. Alicea’s reliance on such materials was consistent with the ordinary practices of law enforcement officers, who “routinely and reasonably rely upon hearsay in reaching their conclusions,” Dukagjini, 326 F.3d at 57.

The expert may not, however, simply transmit that hearsay to the jury. Id. at 54 (“When an expert is no longer applying his extensive experience and a reliable methodology, Daubert teaches that the testimony should be excluded.”). Instead, the expert must form his own opinions by “applying his extensive experience and a reliable methodology” to the inadmissible materials. Id. at 58. Otherwise, the expert is simply “repeating hearsay evidence without applying any expertise whatsoever,” a practice that allows the Government “to circumvent the rules prohibiting hearsay.” Id. at 58-59.

At trial, Alicea was unable to separate the sources of his information, stating that his testimony was based on “a combination of both” custodial interrogations and other sources. On cross-examination, however, Alicea identified hearsay as the source of much of his information. For example, his testimony that the Freeport clique initially funded itself through drug sales was based on “some of the articles that [he] had researched” and “[r]eports from law enforcement personnel.” His testimony about MS-13’s taxation of drug sales by non-members was based on a gang member having told him so during a custodial interrogation in this case. Alicea had learned about MS-13 treasury funds from about a dozen MS-13 members both in and out of custody. Additionally, Alicea discovered his information about MS-13’s involvement in Mexican immigrant smuggling through “research on the Internet,” and more specifically from a website containing a media report and an interview with a law enforcement official. And although Alicea did not identify the source of his statements about the number of firearms the Task Force had seized and the number of murders on Long Island that MS-13 members had committed, we cannot imagine any source for that information other than hearsay (likely consisting of police reports, Task Force meetings, conversations with other officers, or conversations with members of MS-13).

Not all of Alicea’s testimony was flawed, and some of the information that he provided to the jury resulted from his synthesis of various source materials. As a review of his testimony shows, however, at least some of his testimony involved merely repeating information he had read or heard — information he learned from witnesses through custodial interrogations, newspaper articles, police reports, and tape recordings. When asked how he learned particular facts, Alicea did not explain how he had pieced together bits of information from different sources and reached a studied conclusion that he then gave to the jury. Instead, he testified that he had read an article, or had talked to gang members in custody (including, on at least one occasion, a gang member arrested as part of this investigation), or listened to a recording (evidence that could have been played to the jury in its original form, notwithstanding that some informants may have been identified in the process). This testimony strongly suggests that Alicea was acting not as an expert but instead as a case agent, thereby implicating our warning in Dukagjini — a warning the Government appears not to have heard or heeded. Alicea did not analyze his source materials so much as repeat their contents. Alicea thus provided evidence to the jury without also giving the jury the information it needed “to factor into its deliberations the reliability (or unreliability) of the particular source,” Dukagjini, 326 F.3d at 57 n. 7. These statements therefore violated Rule 703.

F. Alicea’s Reliance on Testimonial Statements and Crawford

For similar reasons, some of Alicea’s testimony also violated Crawford. In Crawford, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177, the Supreme Court held that the Confrontation Clause of the Sixth Amendment prohibits the introduction into evidence of the out-of-court testimonial statements made by an absent witness unless that witness is unavailable and the defendant had a prior opportunity for cross-examination. Id. at 54, 124 S.Ct. 1354. While the Court did not provide a comprehensive definition for the term “testimonial,” it placed custodial interrogations within the “core class” covered by the rule it had just announced. Id. at 51, 124 S.Ct. 1354; see also Davis v. Washington, 547 U.S. 813, 822, 126 S.Ct. 2266, 2273, 165 L.Ed.2d 224 (2006) (explaining that a custodial police interrogation after a Miranda warning “ ‘qualifies under any conceivable definition’ of an ‘interrogation’ ” (quoting Crawford, 541 U.S. at 53, 124 S.Ct. 1354)).

When faced with the intersection of the Crawford rule and officer experts, 4 we have determined that an officer expert’s testimony violates Crawford “if [the expert] communicated out-of-court testimonial statements of cooperating witnesses and confidential informants directly to the jury in the guise of an expert opinion.” Lombardozzi, 491 F.3d at 72. As with a Rule 703 challenge to the expert’s reliance on hearsay, the question under Crawford is whether the expert “applied his expertise to those statements but did not directly convey the substance of the statements to the jury,” id. at 73. In fact, when the inadmissible hearsay at issue is a testimonial statement, the Supreme Court has recognized that Rule 703 hearsay claims and Sixth Amendment Crawford claims are “are generally designed to protect similar values.” Dukagjini, 326 F.3d at 56 n. 6 (citing Idaho v. Wright, 497 U.S. 805, 814, 110 S.Ct. 3139, 111 L.Ed.2d 638 (1990)) (remarking that “[i]n this case, the appellants’ hearsay and Confrontation Clause claims are coextensive,” but noting that the Supreme Court has “been careful not to equate” the two types of claims). Because it is a question of law whether an expert witness’s testimony violated Craw ford, our review is de novo. United States v. Wallace, 447 F.3d 184, 186 (2d Cir.2006).

Alicea’s reliance on hearsay is beyond doubt; a more difficult question is the extent to which that hearsay took the form of custodial statements and was thus testimonial. At trial, he testified that he had participated in between fifteen and fifty custodial interrogations of Long Island MS-13 members. He also testified that he had learned through a custodial interrogation that MS-13 taxed non-member drug dealers. The interrogation was one that he conducted as part of the same investigation that resulted in the convictions being appealed here. Among the other facts that he learned at least partially from custodial interrogations were that MS-13 treasury funds were used to purchase narcotics and that MS-13 members used interstate telephone calls to coordinate activities.

We are at a loss in understanding how Alicea might have “applied his expertise” to these statements before conveying them to the jury, such that he could have avoided “convey[ing] the substance of [those] statements to the jury.” Lombardozzi, 491 F.3d at 73. Although the exact source of much of his information remains unclear, there was at least one fact to which Alicea testified — the drug tax — that was based directly on statements made by an MS-13 member in custody (during the course of this very investigation). This impugns the legitimacy of all of his testimony and strongly suggests to us that Alicea was “simply summarizing an investigation by others that [was] not part of the record,” Dukagjini, 326 F.3d at 54, and presenting it “in the guise of an expert opinion,” Lombardozzi, 491 F.3d at 72. We hold, therefore, that Alicea’s reliance on and repetition of out-of-court testimonial statements made by individuals during the course of custodial interrogations violated Appellants’s rights under the Confrontation Clause of the Sixth Amendment.

G. Harmless Error

Having found error in much of Alicea’s testimony, vacatur is required unless we are “convinced that the error was harmless beyond a reasonable doubt.” United States v. Reifler, 446 F.3d 65, 87 (2d Cir.2006). Several factors are relevant when evaluating the error’s likely impact: (1) the strength of the Government’s case; (2) the degree to which the statement was material to a critical issue; (3) the extent to which the statement was cumulative; and (4) the degree to which the Government emphasized the inadmissible evidence in its presentation of its case. Id. Though all of these factors are relevant, we have stated that the strength of the Government’s case is “probably the single most critical factor.” 5 Id.

Alicea’s erroneously admitted statements were relevant to several issues in the case, including whether MS-13 was an enterprise, whether MS-13 “had an effect on interstate or foreign commerce,” whether MS-13 engaged in narcotics trafficking, and whether MS-13 engaged in acts and threats of murder. All of these issues were highly material to the case, so much so that the district judge asked the jury for special findings with regard to each of them. Because the jury found that the Government had failed to show that MS-13 engaged in narcotics trafficking, however, we will not address whether the erroneous admission of expert testimony would have been harmless with respect to that issue.

1. MS~13’s Enterprise Status and Effect on Interstate Commerce

In addition to that portion of Alicea’s testimony that violated either Federal Rule of Evidence 703 or Crawford, the Government introduced a great deal of other evidence relevant to prove MS-13’s status as an enterprise and the organization’s effect on interstate commerce. We are satisfied that Alicea’s erroneously admitted testimony on these issues was harmless beyond a reasonable doubt. For example, on the element of enterprise status, Alicea, in the portion of his testimony addressing MS-13’s background, testified about the gang’s structure, membership rules, symbols, and history; both Appellants, in their confessions, identified themselves as “members” and described MS-13’s membership rules; and the cooperating witnesses testified about MS-13’s narcotics operations, treasury, and membership rules. With respect to the interstate commerce element, Alicea, in his background testimony, testified that MS-13 was a national and international organization with local subunits; Vasquez confessed that the van used during the shootings had been stolen; the cooperating witnesses testified about MS-13’s narcotics operations, the tax imposed on non-member drug dealers, the use of treasury money to purchase firearms, and the gang’s national and international membership, and the use of stolen cars for drive-by shootings; and evidence showed that the firearm used in the shooting had been manufactured outside of New York State. On both elements, Alicea’s erroneously admitted testimony was cumulative, and its admission was harmless.

2. Acts and Threats of Murder

As for proof that the enterprise engaged in acts and threats of murder, however, the Government’s case was weaker and Alicea’s testimony was more material. Though the Government did introduce evidence other than Alicea’s testimony tending to prove that MS-13 engaged in acts and threats of murder, the evidence tending to prove that element was less extensive. Yasquez stated in his confession that “we are at war with the SWP gang.” He further stated that he was threatened with murder during the Hempstead shooting: “They told me to do it, and if I didn’t do that, they would rub me out, meaning kill me.... ” Castro made similar statements in his oral confession, as recounted by a police detective:

Q. Did [Castro] say who the targets of that shooting were?

A. Yes, he said SWP was the target.

Q. Did he say why they were shooting at SWP?

A. He said that they were the enemy, and MS was at war with them.

Q. What if anything did he say about the relationship between MS-13 and the Bloods?

A. He also said MS was at war with the Bloods, and that they, meaning MS, had to protect themselves.

When asked to explain what it means to “make a quota,” Castro stated that it meant “[t]o stab, shoot, beat, kill the enemy, which he described as SWP, the Bloods and 18 Street gangs.” In addition, Admettre testified that MS-13 had a policy “to shoot and kill rival gang members” and “to attack and defend.” He also testified about a separate February 2003 shooting involving both Appellants:

David Vasquez identified [an] SWP member and opened fire. And Castro jumped out along with Vasquez to chase the SWP member down the block and both of them fired.

Menjivar testified that “MS-13 engaged in violent acts such as ‘fights, shootings, beating someone up.’ ”

Despite the volume of evidence that the Government introduced on this element, the question of whether it was harmless to admit Alicea’s testimony — including his statement that MS-13 had committed between eighteen and twenty-three murders on Long Island and his statement that the purpose of MS-13 was to commit murders — remains. On first impression, it is difficult to imagine how a law enforcement officer’s testimony that the gang had killed eighteen to twenty-three people would fail to affect the jury’s finding that the gang engaged in acts and threats of murder. Alicea was alone in testifying that MS-13 had actually committed eighteen to twenty-two or twenty-three murders in the preceding five years. Not considering Ali-cea’s testimony, much of the remaining evidence consists of what is essentially “tough talk,” and none of the remaining evidence shows that MS-13 had committed any murders. Even more so than the evidence of the drive-by shootings and the February 2003 incident, Alicea’s testimony provided proof that this tough talk reflected the gang’s actual practices rather than mere posturing. And while the Government did not spend a great deal of time in its closing argument emphasizing Ali-cea’s testimony on that point, his testimo ny is virtually the first thing that the Government mentioned when directing the jury’s attention to the proof that MS-13 engaged in acts and threats of murder:

You heard from Investigator Alicea, our first witness, that in the five years that he’s been on the task force, the last five years, there have been at least 18 murders here on Long Island, committed by MS 13 members in furtherance of their gang wars.

After carefully considering the evidence introduced at trial, we conclude that the introduction of Alicea’s inadmissible testimony was not harmless beyond a reasonable doubt. The Government was required to prove acts and threats of murder as an element of every offense with which Appellants were charged. Apart from Alicea’s testimony, the Government introduced only circumstantial evidence tending to prove that element; evidence that, though capable of supporting a jury’s finding of guilt, does not compel such a determination. The Government appears to have recognized this possible weakness in its case, as it invoked Alicea’s testimony at the start of its summation discussion, expecting, we assume, that Alicea’s testimonial assertion of eighteen murders would color the jury’s interpretation of the other evidence on this issue. Only in the rarest of cases will we find harmless the admission of the only direct evidence of murder, all other evidence of murderous conduct being partial and inexact. Where, as here, the Government must demonstrate acts and threats of murder, yet the Government introduces no admissible evidence of murder and only circumstantial evidence of threats, we cannot find the admission of direct evidence of multiple murders to have been harmless.

Because proof of the racketeering element, which was based on acts and threats of murder, was essential to securing Appellants’ convictions for conspiracy and assault, and because the assault counts were the predicate offenses for the firearm offenses, our finding of non-harmless error is fatal to the Appellants’ convictions on all counts.

II. Remaining Issues

We have reviewed Appellants’ remaining claims, and we find them to be without merit. Because some of the arguments they raise would arise again in any possible retrial, however, we will address them here.

A. Sufficiency

Appellants argue that the evidence was insufficient to show that MS-13 was an enterprise engaged in racketeering activity. More specifically, they argue a failure to show that MS-13 was an entity that existed separately from the conduct of its members, that MS-13 affected interstate commerce, and that MS-13 engaged in acts and threats of murder. Castro also appears to challenge the sufficiency of the evidence on the element of intent.

On review of a claim of insufficiency of the evidence, we “view the evidence ... in the light most favorable to the government and credit every inference that could have been drawn in its favor.” United States v. Diaz, 176 F.3d 52, 89 (2d Cir.1999). “[T]he convictions must be affirmed, so long as, from the inferences reasonably drawn, the jury might fairly have concluded guilt beyond a reasonable doubt.” Id. (citing Jackson v. Virginia, 443 U.S. 307, 319, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979)). Because “where some government evidence was erroneously admitted, we must make our determination concerning sufficiency taking into consideration even the improperly admitted evidence,” United States v. Cruz, 363 F.3d 187, 197 (2d Cir.2004), we consider the entirety of Alicea’s testimony when evaluating sufficiency.

An enterprise is “a group of persons associated together for a common purpose of engaging in a course of conduct.” United States v. Turkette, 452 U.S. 576, 583, 101 S.Ct. 2524, 69 L.Ed.2d 246 (1981); see also 18 U.S.C. § 1959(b)(2) (defining “enterprise”). This enterprise must have an existence separate from the series of criminal acts that constitute its racketeering activity. Turkette, 452 U.S. at 583, 101 S.Ct. 2529. We have no difficulty finding that the evidence was sufficient to show that MS-13 satisfied this definition. Multiple witnesses described MS-13’s membership rules, organizational treasury, and symbols. In addition, Alicea testified about MS-13’s history, national and international membership, and leadership structure, and Admettre testified that MS-13 had a “policy ... to attack and defend” and that Castro was the leader of the Freeport clique. Drawing all reasonable inferences in favor of the Government, this testimony shows MS-13 to be a national organization, organized into local subunits, with its own leadership structure, membership rules, symbols, policies, and financial operations. Such an organization constitutes an enterprise as defined in Turkette.

With regard to the interstate commerce element, Appellants appear to argue that because the Government relied on narcotics trafficking to show an effect on interstate commerce, the jury’s finding that the Government failed to prove narcotics trafficking as a racketeering activity necessarily means that the Government also failed to show an effect on interstate commerce. The conspiracy and the assault offenses of which Appellants were convicted require that the offense be in furtherance of an enterprise “engaged in, or the activities of which affect, interstate or foreign commerce.” 18 U.S.C. § 1959(b). Transporting goods, such as firearms or stolen vehicles, across state lines is a classic example of engaging in interstate commerce. Cf. United States v. Robertson, 514 U.S. 669, 672, 115 S.Ct. 1732, 131 L.Ed.2d 714 (1995) (“[A] corporation is generally ‘engaged in commerce’ when it is itself ‘directly engaged in the production, distribution, or acquisition of goods or services in interstate commerce.’ ” (quoting United States v. Am. Bldg. Maint. Indus., 422 U.S. 271, 283, 95 S.Ct. 2150, 45 L.Ed.2d 177 (1975))). Use of an instrumentality of commerce, such as telephone lines, is also generally viewed as an activity that affects interstate commerce. Cf. United States v. Atcheson, 94 F.3d 1237, 1243 (9th Cir.1996) (finding defendants’ “placement of out-of-state phone calls” to be a “connection with interstate commerce” under Hobbs Act); United States v. Muskovsky, 863 F.2d 1319, 1325 (7th Cir.1988) (finding effect on interstate commerce based on the use of interstate telephone calls to verify credit card transactions). Beyond these traditional examples, any other conduct having even a de minimis effect on interstate commerce suffices. United States v. Davila, 461 F.3d 298, 306 (2d Cir.2006).

At trial, Alicea testified that out-of-state members had traveled to New York State for MS-13 meetings; that MS-13 treasury funds were used to purchase firearms manufactured outside New York State; that Mexican MS-13 cliques act as smugglers; that MS-13 leaders make interstate telephone calls to coordinate their activities; and that MS-13 is engaged in international narcotics smuggling and the interstate transportation of stolen vehicles. Both Menjivar and Admettre testified that individuals traveled across state lines to attend MS-13 meetings in New York State. Menjivar also testified that MS-13 sent money to individuals located in El Salvador. Based on this evidence, the jury could reasonably have found that MS-13 engages in or affects interstate commerce.

Appellants’ final argument is that the evidence was insufficient to show that MS-13 engaged in acts and threats of murder. In their confessions, both Appellants stated that MS-13 was “at war” with rival gangs. Castro defined making a quota as “[t]o stab, shoot, beat, kill the enemy.” In his confession, Vasquez stated that during the Hempstead shooting, the other individuals in the van “told me to do it, and if I didn’t do that, they would rub me out, meaning kill me.” Moreover, Admettre testified that MS-13 had a policy “to shoot and kill rival gang members.” Menjivar similarly testified that “MS-13 engaged in violent acts such as ‘fights, shootings, beating someone up,’ ” while Alicea — over defense objection — testified that “[t]hey need guns to do what MS-13 does, which is, you know, shoot at rival gang members, and sometimes in the process, obviously, some people get hit.” Finally, Admettre described a February 2003 incident in which Appellants chased and shot at an SWP member. Drawing all reasonable inferences in favor of the Government, and with Alicea’s erroneously admitted testimony taken into consideration, the jury fairly could have found the evidence sufficient to find that MS-13 engaged in acts and threats of murder.

Castro alone further claims that the two assault counts stemming from the Hempstead shootings are multiplicitous because the Government failed to show that either Vasquez (the shooter) or Castro intended to target more than one person. He characterizes the Hempstead shooting as “a single shooting of 4 or 5 shots by another person” that resulted in injuries to two individuals. Because Castro’s challenge hinges on his claim that the Government failed to satisfy the element of intent, we will construe it as a challenge to the sufficiency of the evidence on that issue. We find that the evidence was more than sufficient to find that Castro intended to injure multiple victims. At trial, a police detective described Castro’s confession to the Hempstead shootings:

Q: Did [Castro] say who the targets of that shooting were?

A: Yes, he said SWP was the target.

Q: Did he say why they were shooting at SWP?

A: He said that they were the enemy, and MS was at war with them.

This testimony tends to show that Castro knew that the purpose of the Hempstead shooting was to shoot at multiple people. Furthermore, Menjivar testified that he gave the bullets to Castro, and from that testimony, one can easily infer that Castro knew that the weapon was loaded with multiple bullets. The jury reasonably could have found that Castro intended to harm multiple individuals.

B. Multiplicity of the Indictment

Appellants also claim that the district court should have sentenced them to concurrent sentences for each of the three counts of use of a firearm in furtherance of a crime of violence, 18 U.S.C. § 924(c)(1), because the firearms were all used in furtherance of the same conspiracy. Because § 924(c)(1)(D)(ii) expressly prohibits concurrent sentencing under § 924(c)(1), we construe their challenge as a claim that Appellants’ three § 924(c)(1) convictions were multiplicitous. Therefore, the issue is whether the three underlying shootings constituted separate predicate crimes under § 924(c)(1). Because that issue is a question of law, our review is de novo. United States v. Wallace, 447 F.3d 184, 186 (2d Cir.2006).

In United States v. Lindsay, 985 F.2d 666 (2d Cir.1993), we held that the appro priate unit of prosecution under § 924(c)(1) is the predicate offense (i.e., the “crime of violence”) rather than the number of firearms. Id. at 674. Despite that general rule, we have twice found that multiple predicate offenses gave rise to only a single § 924(c)(1) offense. In the first, United States v. Finley, 245 F.3d 199 (2d Cir. 2001), the defendant had been convicted of two § 924(c)(1) counts. Id. at 201. The predicate crimes were (1) drug possession and (2) drug distribution. Id. Both predicate crimes stemmed from a single transaction in which Finley had (1) possessed multiple packets of cocaine and (2) sold two of those packets to an undercover officer. Id. at 201-02. We reversed Finley’s second § 924(c)(1) conviction, id. at 208, finding that “[t]he statute does not clearly manifest an intention to punish a defendant twice for continuous possession of a firearm in furtherance of simultaneous predicate offenses consisting of virtually the same conduct.” Id. at 207.

We reached a similar conclusion in United States v. Wallace, 447 F.3d 184, where two defendants had each been convicted of two § 924(c)(1) counts in connection with a narcotics-related drive-by shooting. Id. at 185-86. The § 924(c)(1) offenses were based on two separate predicate crimes: (1) conspiracy to traffic in narcotics, and (2) firing of a weapon into a crowd of two or more persons in furtherance of a major drug offense (i.e., the conspiracy). Id. at 187. Despite the existence of two separate predicate crimes, we found that “[t]he relevant conduct underlying the offenses predicating Counts Thirteen and Fourteen consists of the same shooting.” Id. at 189. Both offenses involved one defendant’s use of a firearm in connection with a drug offense, and both involved the murder of a single victim during the drive-by shooting. Id.

We distinguished Finley in United States v. Salameh, 261 F.3d 271 (2d Cir. 2001), where the defendants had each been convicted of multiple crimes in connection with the 1993 bombing of the World Trade Center. Id. at 274. The defendants had each been convicted of two § 924(c)(1) counts. Id. at 275. One count had been based on the predicate crime of conspiracy, and the other count had been based on the predicate crime of assaulting Secret Service officers during the bombing. Id. The defendants there relied on Finley to argue that their two § 924(c)(1) convictions were duplicative. Id. at 277. We found, however, that while the two § 924(c)(1) convictions in Finley were based on a single use of a single firearm, the two § 924(c)(1) counts in Salameh were based on two separate uses of the bomb: transportation and detonation. Id. at 279.

The instant case raises none of the concerns present in Finley and Wallace. We hold, therefore, that the rule in Lindsay, which establishes that the appropriate unit of prosecution under § 924(c)(1) is the predicate offense, applies here. In Finley and Wallace, the defendants had been convicted of various substantive narcotics-related offenses based on single criminal acts. On the other hand, Appellants here were convicted of three counts of the same criminal violation, assault, but the assaults were based on two separate drive-by shootings that resulted in the shooting of three separate victims. As in Salameh, the existence of a single conspiracy connecting the three assaults does not preclude treating them as separate predicate offenses. 261 F.3d at 278-79. And although Appellants now describe the shootings as a “single incident,” the jury found both Appellants culpable of the shooting of three separate individuals. Although those separate shootings are clustered in time and space, that clustering does not somehow merge them into one predicate crime.

This case does not raise the concerns that led us in Finley and Wallace to depart from the general rule in Lindsay. In Finley, we observed that it would be absurd to find multiple § 924(e)(1) offenses when a drug dealer possesses drugs and sells only a portion (thus committing the predicate offenses of possession and distribution) but then to find only a single § 924(c)(1) offense when the same dealer, in possession of the same drugs, sells all of them. 245 F.3d at 208. There, we were responding to the unique manner in which a single course of conduct involving controlled substances, such as continuing possession of a large quantity of contraband, gives rise to multiple possible offenses, such as possession, possession with intent, and so forth. The assaults involved here raise no concerns of that sort, and thus Finley and Wallace do not control the outcome.

C. Prior Act Evidence

Appellants also object to the district court’s admission of evidence relating to an uncharged shooting and of evidence that MS-13 engaged in narcotics trafficking. They claim that this evidence was offered to show their propensity for criminality and that it was overly prejudicial. With respect to the evidence of narcotics trafficking, they argue that it should have been excluded because it was likely to cause the jury to reach a verdict on an improper basis.

Evidence of uncharged crimes is inadmissible when offered to show a person’s propensity to act in a particular way. Fed.R.Evid. 404(b). Such evidence is admissible, however, when offered to show “motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident.” Id. Given our “inclusionary approach,” prior act evidence is admissible if offered “for any purpose other than to show a defendant’s criminal propensity.” United States v. Garcia, 291 F.3d 127, 136 (2d Cir.2002) (internal quotation marks omitted). We review the district court’s admission of prior act evidence for abuse of discretion, and the district court’s ruling stands unless it was arbitrary and irrational. United States v. Lombardozzi, 491 F.3d 61, 78-79 (2d Cir. 2007).

First, Appellants’ arguments regarding evidence of narcotics trafficking have little merit. MS-13’s involvement in narcotics trafficking was an element of the charged offenses, and as a result, evidence of narcotics trafficking is not prior act evidence, but rather direct evidence of the charged offense. Second, with regard to evidence of the February 2003 shooting, we find no error. The district court admitted the evidence to show the existence of the racketeering enterprise, and it expressly instructed the jury as to the limited use it could make of that evidence. Even the defense counsel stated that “I don’t dispute that one could conceive the testimony that the government wants to adduce as being background proof of the existence of a conspiracy and 404(b).” Where, as here, the existence of a racketeering enterprise is at issue, evidence of uncharged crimes committed by members of that enterprise, including evidence of uncharged crimes committed by the defendants themselves, is admissible “to prove an essential element of the RICO crimes charged — the existence of a criminal enterprise in which the defendants participated.” Matera, 489 F.3d at 120 (upholding admission of evidence of uncharged murders). The evidence was also admissible to show the existence of the conspiracy with which both Appellants were charged. United States v. Diaz, 176 F.3d 52, 79 (2d Cir.1999) (“Where ... the indictment contains a conspiracy charge, uncharged acts may be admissible as direct evidence of the conspiracy itself.” (internal quotation marks omitted)). Although Appellants are correct in noting that evidence of a prior shooting presents a significant risk of prejudice, “[w]hen a defendant engages in a criminal enterprise which involves very serious crimes, there is a likelihood that evidence proving the existence of the enterprise through its acts will involve a considerable degree of prejudice. Nonetheless, the evidence may be of important probative value in proving the enterprise.” Matera, 489 F.3d at 121. Under these circumstances, the district court did not abuse its discretion when it admitted the evidence. 6

D. Constructive Amendment of the Indictment

Appellants also argue that because the Indictment alleged that MS-13 engaged in two forms of racketeering activity (acts and threats of murder, and narcotics trafficking), and because the jury did not find that MS-13 engaged in narcotics trafficking, “the specific enterprise set out in the indictment was not proven.” This claim is entirely without merit and requires little discussion: “Where there are several ways to violate a criminal statute ... federal pleading requires ... that an indictment charge [be] in the conjunctive to inform the accused fully of the charges. A conviction under such an indictment will be sustained if the evidence indicates that the statute was violated in any of the ways charged.” United States v. McDonough, 56 F.3d 381, 390 (2d Cir.1995) (internal quotation marks, citations, and alterations omitted). Thus, although the Indictment alleged that MS-13 engaged in both acts and threats of murder and narcotics trafficking, Appellants’ “conviction under [that] indictment” was proper because the jury found that “the evidence indicate[d] that the statute was violated in [the other] way[ ] charged.” Id.

E. Sentencing Under § 924(c)(1)

Appellants’ final claim is that the district court erred when it imposed 25-year sentences for their second and third § 924(c)(1) offenses based on 18 U.S.C. § 924(c)(1)(C)®, which requires a 25-year sentence for each “second or subsequent conviction” under § 924(c)(1). They argue that because the Indictment did not charge that any of the firearm counts were “second or subsequent” convictions, the jury did not find Appellants guilty of having committed any second or subsequent firearm offenses, and its verdict therefore did not support such a finding by the district court. This claim is foreclosed by United States v. Campbell, 300 F.3d 202 (2d Cir. 2002), where we held that because the fact of a prior conviction is not an element of § 924(c), and because Apprendi v. New Jersey, 530 U.S. 466, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000), excluded prior convictions from its requirements, “there was no requirement that the existence of prior convictions be alleged in the indictment or that any of the multiple firearms convictions returned by the jury be described by the jury as second or subsequent.” 7 Campbell, 300 F.3d at 212-213. Campbell is directly on point and controls our review here; the sentences imposed by the district court did not violate Apprendi.

CONCLUSION

Because the testimony of the Government expert witness violated the Federal Rules of Evidence and the Confrontation Clause of the Sixth Amendment, and because that error was not harmless, we VACATE Appellants’ convictions on all counts and REMAND to the district court in the event that the Government chooses to retry the case.

1 . MS-13 has also appeared in cases in the Fourth, Fifth, Sixth, and Eleventh Circuits. See United States v. Calles, 250 Fed.Appx. 939 (11th Cir.2007); United States v. Funes, 248 Fed.Appx. 593 (5th Cir.2007); United States v. Hernandez-Villanueva, 473 F.3d 118 (4th Cir. 2006); Castellano-Chacon v. INS, 341 F.3d 533 (6th Cir.2003).

2 . The Government’s theory was that the ammunition that was used to commit the drive-by shootings constituted an "explosive” under federal law.

3 . After initially characterizing how he learned about the drug tax as a casual conversation with an MS-13 member, Alicea later clarified that he learned about the drug tax during a custodial interrogation of an MS-13 member. Upon further questioning, Alicea stated that this "conversation" had taken place "at the United States Attorney’s Office.” The MS-13 member with whom Alicea had been "conversing” had been indicted and had not been released on bail, and he had been escorted to the "conversation” by "the U.S. marshals.” When asked why that MS-13 member had been arrested, Alicea answered that the individual was "[p]art of this investigation as well, yes.”

In other words, Alicea learned about the drug tax from an MS-13 member in custody during the course of this very investigation. The interrogation took place at the U.S. Attorney’s Office. The record does not reflect whether the Assistant United States Attorney in charge of this case was present.

4 . We have not been alone in facing questions about the collision between Crawford and testimony by police officers. As the First Circuit has noted, ”[p]ost-Crawford, the admission of non-testifying informants' out-of-court testimonial statements, through the testimony of police officers, is a recurring issue in the courts of appeals.” United States v. Maher, 454 F.3d 13, 19 (1st Cir.2006).

5 . Although harmless error analysis originally focused on whether the error had affected the jury, see Chapman v. California, 386 U.S. 18, 23-24, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967) (defining an error as harmless only where there was no "reasonable possibility that the evidence complained of might have contributed to the conviction"), over the years our focus has shifted from the impact of the error on the jury's analysis to an assessment of the strength of the remaining evidence of guilt, see Stephen Alan Childress & Martha S. Davis, 2 Federal Standards of Review § 7.03 (3rd ed.2004) (identifying this shift in Harrington v. California, 395 U.S. 250, 254, 89 S.Ct. 1726, 23 L.Ed.2d 284 (1969), where the Supreme Court determined the harmlessness of an error by evaluating whether the properly admitted evidence of guilt was "overwhelming”); Schneble v. Florida, 405 U.S. 427, 432, 92 S.Ct. 1056, 31 L.Ed.2d 340 (1972) (holding error harmless because "the minds of an average jury would not have found the State’s case significantly less persuasive” in the absence of the erroneously admitted evidence (internal quotation marks omitted)). That shift, away from a retrospective reconstruction of the actual jury’s deci- sionmaking process and toward a metajuridical evaluation of the importance the error should reasonably have had, has given rise to multiple types and levels of uncertainty. On the general level of the structure of our review, the shift creates "uncertainty ... [as to] whether the reviewing court is to consider the effect of the error on the jury or predict what verdict would have been rendered in the absence of the error.” Peck v. United States, 102 F.3d 1319, 1326 (2d Cir.1996) (en banc) (Newman, J., concurring).

On the more specific level of administering the multifactored balancing tests that we have developed in this context, the question arises whether the reviewing court must conduct its review having drawn all reasonable inferences from the remaining evidence in favor of the defendant. See Schneble, 405 U.S. at 435, 92 S.Ct. 1056 (Marshall, /., dissenting) ("[A]ll reasonable inferences that might be drawn from the evidence must be drawn in favor of the defendant, since the jury may very well have made just these inferences.”); United States v. Santos, 449 F.3d 93, 100 (2d Cir. 2006) (reviewing government’s principal evidence in light of what "reasonable jurors could have determined”). We have never held that drawing those inferences is part of harmless error review, however, and prevailing precedent dictates that it is not. See Rose v. Clark, 478 U.S. 570, 579, 106 S.Ct. 3101, 92 L.Ed.2d 460 (1986) ("Where a reviewing court can find that the record developed at trial establishes guilt beyond a reasonable doubt, the interest in fairness has been satisfied and the judgment should be affirmed.” (emphasis added)); United States v. Reifler, 446 F.3d 65, 88-90 (2d Cir.2006) (omitting any mention of reasonable inferences when articulating the factors relevant to harmless error review); Zappulla v. New York, 391 F.3d 462, 468 (2d Cir.2004) (same).

Because the result in this case is the same whether or not we draw inferences in favor of the defendants, we need not resolve any such conflict here. We take this opportunity, however, to flag for the Court our concern that our approach to harmless error analysis proceed consistently.

6 . The district court's failure to balance the probative value of the evidence with the possibility of prejudice on the record does not change this conclusion. Defense counsel objected to admission of the evidence because "the probative value here [does not] outweigh ] the prejudice to this jury.” When the district court admits evidence after hearing such an objection, it is "not required to make mechanical recitation that it balanced probative value against prejudicial effect”; the balancing is presumed. United States v. Pitre, 960 F.2d 1112, 1120 (2d Cir.1992).

7 . Despite the striking parallels between Campbell and the facts here, neither Appellant cited to or acknowledged Campbell when making their Apprendi claims.