THACKER, Circuit Judge.

William White ("Appellant") believed his ex-wife ("MW") owed him money. When she refused to pay, he sent her a series of e-mails, four of which threatened violence if MW did not meet his demands. MW reported the threats to the authorities, and Appellant was eventually charged in a four-count indictment with violating 18 U.S.C. § 875(b), which makes it a felony to transmit threats in interstate commerce with the intent to extort. After trial, a jury convicted him of three of the charged § 875(b) counts, and one count of the lesser-included offense of transmitting a threat (without the intent to extort), in violation of § 875(c). The district court sentenced Appellant to a 92-month term of imprisonment.

Appellant now asks us to reverse his conviction and vacate his sentence, assigning a number of errors. He maintains he could not have intended to extort MW because she owed him a legitimate debt and alleges more generally that the district court misinstructed the jury on the mens rea requirements for conviction pursuant to § 875(b) and (c). He also complains that the use of an anonymous jury at his trial was improper; asserts that the district court erroneously admitted hearsay evidence; challenges the sufficiency of the evidence presented against him; and disputes both the procedural and substantive reasonableness of his sentence. We are not persuaded that any of Appellant's arguments undermine the jury's verdict or the district court's sentence. We therefore affirm the district court's judgment for the reasons that follow.

I.

A.

This is not Appellant's first brush with the law for making threats, and his prior misadventures set the stage for this case. In 2010, he was charged in the Western District of Virginia for making a threatening telephone call to a university administrator and sending intimidating letters to several tenants in Roanoke who had filed a fair housing complaint against their landlord. A jury convicted him, and the district court imposed a 30-month term of imprisonment.

While he was incarcerated, Appellant's relationship with his now-ex-wife, MW, deteriorated. They eventually separated and MW agreed1 to pay alimony to Appellant. She made the first two payments in March and April of 2012. Around the same time, in March 2012, we upheld Appellant's conviction on appeal, but remanded the case for resentencing. See United States v. White, 670 F.3d 498, 502-03, 515-16 (4th Cir.2012).

Appellant was out of prison and on supervised release by that time, so the district court set a resentencing hearing for May 14, 2012. Appellant didn't show. Instead, he fled, absconding to Mexico with an acquaintance named Sabrina Gnos. When MW learned Appellant was on the lam, she stopped making the alimony payments, at least in part because she feared that doing so would amount to aiding a fugitive. Appellant's subsequent attempts to persuade MW to resume making the payments form the basis of the indictment in this case.

Appellant sent MW the following messages, which form Counts I, II, and III of the indictment at issue here, between May 27th and May 29th:

J.A. 17-18.2

Appellant also asked Gnos for help finding someone in Virginia to pressure MW into making the payments. On June 2, 2012, Gnos, who was by that time cooperating with the Federal Bureau of Investigation ("FBI"), recorded the following conversation:

J.A. 730. The following day brought more of the same:

Id. at 733. Appellant followed up again on June 4th:

Id. at 737.

Finally, on June 7, 2012, after Gnos failed to find a solution, the indictment alleges Appellant sent MW a final warning, charged in Count IV:

J.A. 18.

The following day, June 8, 2012, Appellant was arrested in Mexico and eventually deported to the United States. On February 7, 2013, he was indicted in the Western District of Virginia and charged with four counts of violating 18 U.S.C. § 875(b) on the basis of the e-mails set forth above. As relevant here, the statute penalizes "[w]hoever, with intent to extort from any person ... any money ..., transmits in ... foreign commerce any communication containing any threat to ... injure the person of another[.]" 18 U.S.C. § 875(b).

B.

Before trial, Appellant moved to dismiss the indictment, arguing that it failed as a matter of law because he had a legal right to the alimony payments he demanded. The district court denied the motion. At the Government's request, and over Appellant's objection, the case proceeded to trial before an anonymous jury.

At trial, Gnos testified in detail about Appellant's activities during his escape to Mexico. She noted that Appellant frequently used a Toshiba laptop during the trip and that Appellant explained he was using software to disguise the computer's Internet Protocol ("IP") address. Gnos also testified that, after she returned to Virginia, she continued to communicate with Appellant, check his mail, and wire him money. At her father's urging, Gnos explained, she eventually contacted authorities and agreed to record her telephone conversations with Appellant. The Government played several of those recordings for the jury, including the clips from early June, described above, in which Appellant repeatedly asked Gnos to find someone to lean on MW so that she would resume making the disputed alimony payments. The court also received into evidence a handwritten note Gnos made of a call with Appellant, which had not been recorded. According to Gnos's note, the tenor of the call was much like the others; Appellant told her, "OK — if phone call dont [sic] work — we will have to have someone fuck her up!" J.A. 727.

Other testimony implicated Appellant as the author of the e-mails charged in the indictment. FBI Agent David Church testified that the e-mails to MW originated from an e-mail address, dhyphen@yahoo. com, that Appellant had previously used. Church also explained that Appellant bragged on his Facebook account of using an IP anonymizer much like the technology Gnos testified Appellant had described to her. And Church testified that Appellant's Facebook account registered activity very near in time to the moments when threatening e-mails were sent to MW, and that both the Facebook activity and e-mails originated from the same (albeit anonymized) IP address.

MW testified that the e-mails made her fearful for her safety and the safety of her daughter. For example, she testified that, after receiving the May 27, 2012 e-mail, she went to the local police station to ask for a protective order. She also shared the e-mails with the United States Marshals Service and the FBI, and took care not to travel alone whenever possible until Appellant had been captured.

Appellant took the stand in his own defense. He testified he did not send MW any of the e-mails identified in the indictment, suggesting instead that Gnos was responsible. The jury deliberated for just over three hours, but ultimately rejected Appellant's version of events, finding him guilty of violating § 875(b) when he sent the May 27th, May 29th, and June 7th e-mails; the jury also convicted Appellant of the lesser-included-offense of violating § 875(c) on the basis of the May 28th e-mail.

The Probation Department thereafter prepared a Presentence Report ("PSR") that recommended a sentence of 92-115 months, based on a total offense level of 26 and a criminal history category of IV. Appellant's offense level was enhanced by two points for obstruction of justice as a result of his trial testimony, and the PSR opined that Appellant's counts of conviction were not subject to grouping pursuant to § 3D1.2(d) of the United States Sentencing Guidelines ("Guidelines"). Appellant objected to the enhancement and also requested a downward departure from the recommended Guidelines sentence. The district court denied Appellant's objection and sentenced him to a 92-month term of imprisonment, at the low end of the Guidelines range.

II.

Appellant raises several issues on appeal, attacking each stage of his prosecution. The heart of his appeal, however, concerns the legal requirements for conviction pursuant to § 875(b) and (c). Appellant claims the indictment against him was legally deficient (and therefore should have been dismissed) because he could not have intended to "extort" alimony payments to which he was legally entitled. He also maintains the jury was improperly instructed on the intent elements of § 875(b) and (c). We review both issues de novo. See United States v. Said, 798 F.3d 182, 193 (4th Cir.2015) ("We review de novo a district court's denial of a motion to dismiss an indictment where the denial depends solely on questions of law."); United States v. Jefferson, 674 F.3d 332, 351 (4th Cir.2012) ("We review de novo the claim that a jury instruction failed to correctly state the applicable law.").

Sections 875(b) and 875(c) both prohibit transmitting "in interstate or foreign commerce any communication containing any threat to kidnap any person or any threat to injure the person of another." 18 U.S.C. § 875(b) and (c). The distinction between the two is that § 875(b) is violated only when such a threat is transmitted with the specific intent to extort something of value, whereas § 875(c) says nothing about the speaker's intent. Appellant's challenges to his § 875(b) convictions turn on the meaning of "intent to extort." His appeal of his § 875(c) conviction, which we consider first, depends on the application of the Supreme Court's recent decision in Elonis v. United States, ___ U.S. ___, 135 S.Ct. 2001, 192 L.Ed.2d 1 (2015).

A.

Section 875(c)

1.

We have previously held § 875(c) is violated if (1) the defendant knowingly communicates a statement in interstate commerce that (2) contains a "true threat" that is not protected by the First Amendment. See United States v. White, 670 F.3d 498, 508-10 (4th Cir.2012) (discussing our prior cases and collecting those of other circuits). And we have explained that a "true threat" in the constitutional sense is one that a reasonable recipient who is familiar with the circumstances would interpret as a serious expression of an intent to do harm. See id. But because the text of § 875(c) does not articulate any additional intent requirements, we have repeatedly held that neither the statute nor the Constitution requires the Government to prove that a defendant subjectively intended the recipient of the communication to understand it as threatening. See, e.g., United States v. Darby, 37 F.3d 1059, 1066 (4th Cir.1994); White, 670 F.3d at 509-11. That is how the district court instructed the jury in this case. Appellant maintains that to do so was error in light of the Supreme Court's decision in Elonis.

In Elonis the Supreme Court asked "whether [§ 875(c)] ... requires that the defendant be aware of the threatening nature of the communication, and — if not — whether the First Amendment requires such a showing." Elonis, 135 S.Ct. at 2004. The Court began by explaining that the text of § 875(c) "requires that a communication be transmitted and that the communication contain a threat," but acknowledged, as we have previously observed, that no "mental state with respect to th[ose] elements" is otherwise specified. Id. at 2008. The Court nevertheless explained that the "`mere omission from a criminal enactment of any mention of criminal intent' should not be read `as dispensing with it.'" Id. at 2009 (quoting Morissette v. United States, 342 U.S. 246, 250, 72 S.Ct. 240, 96 L.Ed. 288 (1952)). Instead, because courts "generally interpret[] criminal statutes to include broadly applicable scienter requirements, even where the statute by its terms does not contain them," id. (internal quotation marks and citations omitted; alteration in original), the Court found it appropriate to "read into the statute only that mens rea which is necessary to separate wrongful conduct from otherwise innocent conduct," id. at 2010 (internal quotation marks omitted).

Turning to § 875(c), the Court emphasized "the crucial element separating legal innocence from wrongful conduct is the threatening nature of the communication." See Elonis, 135 S.Ct. at 2011 (internal quotation marks omitted). As a result, the Court concluded, a defendant may not be convicted of violating § 875(c) based "solely on how his [words] would be understood by a reasonable person," because doing so would impermissibly allow "criminal liability" to "turn solely on the results of an act without considering the defendant's mental state." See id. at 2011-12. Instead, the Court held that to violate § 875(c), a defendant must transmit "a communication for the purpose of issuing a threat, or with knowledge that the communication will be viewed as a threat," or, possibly, with reckless disregard for the likelihood that the communication would be perceived as a threat. Id. But, importantly, the Court's holding in Elonis was purely statutory; and, having resolved the question on statutory grounds, the Court declined to address whether a similar subjective intent to threaten is a necessary component of a "true threat" for purposes of the First Amendment. See id. at 2012 ("Given our disposition, it is not necessary to consider any First Amendment issues.").

Thus, Elonis abrogates our prior holding that liability under § 875(c) can turn solely on how a recipient would interpret a statement, without regard to whether the speaker intended it as a threat. Contra White, 670 F.3d at 508 ("[B]ecause the threat element is not part of the mens rea, it becomes an element of the crime that must be established without consideration of the defendant's intent."). But Elonis does not affect our constitutional rule that a "true threat" is one that a reasonable recipient familiar with the context would interpret as a serious expression of an intent to do harm. See White, 670 F.3d at 508-10.

What that means, in this circuit after Elonis, is that a conviction pursuant to § 875(c) now entails "what the [statute requires] (a subjectively intended threat) and [also] what constitutional avoidance principles demand (an objectively real threat)." See United States v. Jeffries, 692 F.3d 473, 485 (6th Cir.2012) (Sutton, J., dubitante). That is: (1) that the defendant knowingly transmitted a communication in interstate or foreign commerce; (2) that the defendant subjectively intended the communication as a threat; and (3) that the content of the communication contained a "true threat" to kidnap or injure. To prove the second element, the Government, consistent with Elonis, must establish that the defendant transmitted the communication "for the purpose of issuing a threat, or with knowledge that the communication will be viewed as a threat," or, perhaps, with reckless disregard for the likelihood that the communication will be viewed as a threat. See Elonis, 135 S.Ct. at 2012-13. And to establish the third element, in keeping with our prior cases, the prosecution must show that an ordinary, reasonable recipient who is familiar with the context in which the statement is made would interpret it as a serious expression of an intent to do harm. See White, 670 F.3d at 508-10.

Here, by contrast, the district court (which did not have the benefit of the Court's decision in Elonis) instructed the jury that it could convict Appellant pursuant to § 875(c) if he transmitted a true threat in interstate commerce, without regard to his subjective intent. In light of Elonis, that instruction was erroneous. See United States v. Houston, 792 F.3d 663, 667 (6th Cir.2015).

2.

The Government nevertheless maintains the error was harmless. We agree.

The Supreme Court has "often applied harmless-error analysis to cases involving improper instructions," Neder v. United States, 527 U.S. 1, 9, 119 S.Ct. 1827, 144 L.Ed.2d 35, (1999), as have we, see United States v. Brown, 202 F.3d 691, 699 (4th Cir.2000). "[O]ur task is to determine whether the guilty verdict actually rendered [at] trial was surely unattributable to the error." Brown, 202 F.3d at 699 (internal quotation marks omitted). To do so, we ask whether it is "clear beyond a reasonable doubt that a rational jury would have found the defendant guilty absent the error[.]" Neder, 527 U.S. at 18, 119 S.Ct. 1827.

Where, as here, the district court declines to give an instruction "not required under precedent that the Supreme Court later supersede[s]," United States v. Ramos-Cruz, 667 F.3d 487, 496 (4th Cir. 2012), we engage in two specific inquiries to test the harmlessness of the omission. Under the first, we will find an erroneous instruction harmless if we "conclude[] beyond a reasonable doubt that the omitted element was uncontested and supported by overwhelming evidence." Neder, 527 U.S. at 17, 119 S.Ct. 1827. Under the second, in cases where the defendant has contested the omitted element, we ask "whether the record contains evidence that could rationally lead to a contrary finding with respect to that omitted element. If not, then the error was harmless. If so, however, reversal is necessary." Ramos-Cruz, 667 F.3d at 496 (internal quotation marks and citations omitted).

The omitted issue in this case is whether Appellant sent the e-mail charged in Count II of the indictment for the purpose of issuing a threat, with the knowledge that the communication would be viewed as a threat, or (perhaps) with reckless disregard that the e-mail would be perceived as threatening.3 Appellant did not suggest that he sent any of the e-mails in question as a joke, nor did he testify that he was simply blowing off steam. Instead, he contested the issue of intent, at best and if at all, only implicitly — by denying that he sent the e-mails. The jury, however, resoundingly rejected that theory. Thus, the jury having concluded beyond a reasonable doubt that Appellant sent the e-mails, we are left to consider only whether the contents of the Count II e-mail, in the absence of any alternative explanation from Appellant, would permit a jury to rationally find that Appellant did not intend the message as a threat or know it would be received as a threat. We think no rational jury could reach that conclusion.

We acknowledge that appellate courts are ill-equipped to "evaluate states of mind based on a cold record." See Houston, 792 F.3d at 669. And we appreciate that, in many threat cases, the question of intent will be far from clear, even where it is undisputed that the defendant transmitted the communication and the message itself contains harsh and inflammatory language. In Elonis, for example, the defendant posted threatening language on his own Facebook page with disclaimers that the posts were rap lyrics, suggesting that they may have been created for personal, therapeutic, rather than malevolent, reasons. See 135 S.Ct. at 2004-05. Similarly, the Sixth Circuit recently declined to find harmless error in a § 875(c) prosecution where a defendant threatened his lawyer in a call to a relative, rather than to the lawyer himself, because the circumstances of the call made it plausible that the defendant was simply "venting his frustration to a trusted confidante rather than issuing a public death threat to another." See Houston, 792 F.3d at 667-68.

But this case, and the message at issue, give rise to no comparably reasonable competing inferences. The jury found that Appellant sent the e-mail constituting Count II directly to MW, the intended recipient. And the language used — "you are going to have the living shit beat out of you — to start with" — was direct and declarative, not circumspect or hypothetical. Appellant offered no other explanation for the message. And in his contemporaneous conversations with Gnos, he explained his desire to scare MW into resuming the alimony payments.

"In many cases, the predicate facts conclusively establish intent, so that no rational jury could find that the defendant committed the relevant criminal act but did not intend to cause injury." Pope v. Illinois, 481 U.S. 497, 503, 107 S.Ct. 1918, 95 L.Ed.2d 439 (1987) (internal quotation marks omitted). That holds true here. Accordingly, because the record contains no evidence that could rationally lead a jury to conclude that the sender of the Count II e-mail intended to do anything other than threaten the recipient, and because the jury concluded beyond a reasonable doubt that Appellant was indeed the sender, the district court's instructional error was harmless.

B.

Section 875(b)

We now turn to Appellant's three § 875(b) convictions. Like § 875(c), that statute prohibits the transmission in interstate or foreign commerce of threats to kidnap or injure. But, as noted, § 875(b) also requires the threatening communication to be sent "with intent to extort from any person, firm, association, or corporation, any money or other thing of value[.]" 18 U.S.C. § 875(b). Appellant maintains the district court erred by failing to instruct the jury, consistent with Elonis, that he intended the e-mails charged in Counts I, III, and IV as extortionate threats, or knew or recklessly disregarded that those e-mails would be perceived as extortionate threats. He also claims the indictment should have been dismissed because he could not have intended to "extort" from MW alimony payments to which he was legally entitled. The meaning of "intent to extort," however, forecloses both arguments.

1.

Section 875 does not define "intent to extort" or even the term "extortion." See United States v. Coss, 677 F.3d 278, 283-84 (6th Cir.2012) ("The precise meaning of `extort' ... in the context of 18 U.S.C. § 875(d) is an issue of first impression in the Sixth Circuit."); United States v. Jackson, 180 F.3d 55, 65 (2d Cir.1999) (noting that § 875(d) "does not define the terms `extort' or `intent to extort'"). However, two of our sister circuits have persuasively reasoned, and we agree, that § 875 employs "the traditional concept of extortion, which includes an element of wrongfulness." Jackson, 180 F.3d 55, 70-71 (2d Cir.1999); Coss, 677 F.3d 278, 285 (6th Cir.2012) (importing into § 875 "the broader concept of extortion, which carries with it the use of a wrongful threat to procure something of value" (emphasis in original)).

Incorporating this understanding, we hold that the intent to extort for purposes of § 875(b) is the intent to procure something of value through the use of a wrongful threat to kidnap or injure the person of another. Such a threat is wrongful when delivered intentionally. Cf. Elonis, 135 S.Ct. at 2012-13. But this helps Appellant not at all, because it would be passing strange, indeed impossible, for a defendant to intend to obtain something by communicating such a threat without also intending, understanding, or, possibly, recklessly disregarding that the communication would be perceived as threatening, as Elonis requires. The reason is straightforward: Extortion only works if the recipient of the communication fears that not paying will invite an unsavory result. Thus, to intend to extort one must necessarily intend to instill fear of harm (for purposes of § 875(b), in the form of kidnapping or physical injury). Cf., e.g., United States v. Marsh, 26 F.3d 1496, 1500-01 (9th Cir.1994) ("For attempted extortion,... the victim's state of mind is not important. What is important is that the defendant attempted to instill fear in the victim.") (internal quotation marks omitted); United States v. Frazier, 560 F.2d 884, 887 (8th Cir.1977) ("Proof of an attempt to arouse fear is sufficient proof of an attempted extortion under the Hobbs Act."). In other words, the intent to carry out an unlawful act by use of a threat necessarily subsumes the intent to threaten.

The Ninth Circuit's decision in United States v. Stewart, 420 F.3d 1007 (2005), illustrates well the principle we have just articulated. There the court considered whether 18 U.S.C. § 115(a)(1)(B) — which makes it a crime to threaten certain federal officials with the intent to impede, intimidate, interfere with, or retaliate against the official in the performance of his or her duties — requires proof that the defendant subjectively intended his words to be threatening. As the court explained, "proof that the speaker intended the speech to impede, intimidate, interfere with, or retaliate against the protected official" necessarily includes the subjective intent to threaten, because "one cannot have the intent" to impede, intimidate, interfere, or retaliate through the use of a threat "without also intending to make the threat." See id. at 1017, 1019. Likewise, here, we conclude that one cannot have the intent to scare someone into relinquishing property or something of value by communicating a wrongful threat to kidnap or injure without also intending the communication to be threatening.

The district court's instruction sufficiently captured this concept. As the court explained, to convict Appellant pursuant to § 875(b), the jury was required to find that he acted "with intent to ... wrongfully induce someone to pay money or something of value by threatening to injure that person if such payment is not made." J.A. 1076. Accordingly, because the charge required the jury to find that Appellant intended to threaten MW to induce her to pay the disputed alimony, we find no error in the instruction.

2.

Appellant's remaining argument — that he could not have intended to extort MW because he had a "claim of right" to the alimony payments — is similarly unavailing. As we have now said, the question is whether Appellant intended to procure something of value from MW through the use of a wrongful threat to kidnap or injure. The key word is "wrongful."

Appellant's argument is that it is not wrongful to demand money one is rightfully owed. There are situations where that may well be true. In extortion cases under the Hobbs Act, for example, courts have held "a defendant cannot be found guilty of wrongfully obtaining property through the use of a legitimate economic threat if he has a claim of right to the property." See, e.g., United States v. Sturm, 870 F.2d 769, 773 (1st Cir.1989) (emphasis supplied). Instead, "the use of legitimate economic threats to obtain property is wrongful only if the defendant has no claim of right to that property." Id. (footnote omitted).

But this case involves threats of violence, not "legitimate economic threats." And, outside the context of labor relations,4 the "claim of right" defense is inapplicable in Hobbs Act cases involving the use or threatened use of violence. See, e.g., United States v. Villalobos, 748 F.3d 953, 956 (9th Cir.2014) ("[T]he claim of right defense ... is unavailable in cases involving physical violence ... because such violence is inherently wrongful."). This is so, courts have explained, because "Congress meant to punish as extortion" under the Hobbs Act "any effort to obtain property by inherently wrongful means, such as force or threats of force ..., regardless of the defendant's claim of right to the property." United States v. Zappola, 677 F.2d 264, 268-69 (2d Cir.1982) (emphasis supplied).

It makes some sense, then, that the small number of courts to incorporate the "claim of right" defense into their understanding of § 875 have done so in the context of § 875(d), which prohibits the transmission of extortionate threats to the property or reputation of the recipient, rather than threats to kidnap or injure. See 18 U.S.C. § 875(d). As with the Hobbs Act cases, the defense has been limited in the § 875(d) context to a narrow category of threats to reputation deemed not "inherently wrongful." In United States v. Jackson, for example, the Second Circuit opined that it would not be wrongful for a club to threaten to publish the names of members with delinquent accounts if the dues were indeed owing, or for a consumer to register a public complaint about the quality of a seller's product if the product was actually defective. See 180 F.3d at 70-71 ("[I]f the club posts a list of members with unpaid dues and its list is accurate, the dues generally will be paid; if the consumer lodges her complaint and is right, she is likely to receive her refund; and both matters are thereby concluded."). On the other hand, the Jackson court expressly distinguished "extortionate threats to kidnap or to injure a person," explaining that such "conduct ... plainly is inherently wrongful." Id. at 67; see also Coss, 677 F.3d at 284 (quoting Jackson, 180 F.3d at 67). And no court has suggested, as Appellant does, that such threats of physical violence would cease to be wrongful simply because a legitimate debt is at issue.

In sum, just as "you cannot beat someone up to collect a debt, even if you believe he owes it to you," United States v. Castor, 937 F.2d 293, 299 (7th Cir.1991) (internal quotation marks omitted), it follows that a defendant may not threaten to injure or kidnap a person to collect a debt, even one legitimately due and owing. The indictment in this case alleged that Appellant threatened to have MW beaten, hospitalized, or subjected to some less specific violence. Accordingly, even assuming MW owed Appellant the alimony payments he sought, Appellant was not entitled to have the indictment against him dismissed on the basis of his "claim of right" theory.

III.

Having resolved the legal framework, we now consider, and reject, Appellant's remaining objections to the proceedings below.

A.

The Anonymous Jury

Appellant claims the district court erred by empaneling an anonymous jury, a decision which we review for abuse of discretion. See United States v. Hager, 721 F.3d 167, 186 (4th Cir.2013).

A district court should rarely empanel an anonymous jury, but may do so if "(1) there is strong reason to conclude that the jury needs protection from interference or harm, or that the integrity of the jury's function will be compromised absent anonymity; and (2) reasonable safeguards have been adopted to minimize the risk that the rights of the accused will be infringed." Hager, 721 F.3d at 186 (4th Cir.2013) (quoting United States v. Dinkins, 691 F.3d 358, 372 (4th Cir.2012)).

In assessing the need to protect the jury and its functions, a district court should consider several factors, including: whether the defendant is involved in organized crime or a member of some other group with the capacity to harm jurors; whether he has previously attempted to interfere with the judicial process; whether he is facing a lengthy sentence or substantial fine; and whether extensive publicity makes it more likely that the jury will be subjected to intimidation or harassment. See Hager, 721 F.3d at 187. As is often true of multi-factor tests, however, the list is not exhaustive and the presence or absence of any of those is not dispositive. See id.

Here, the district court found Appellant's criminal history weighed heavily in favor of jury anonymity. As the district court carefully explained, Appellant's previous convictions each "reflect his willingness to use threats or personal information to intimidate persons involved in judicial proceedings." J.A. 145. For example, his 2010 conviction in the Western District of Virginia involved threats intended to influence, delay, or prevent the testimony of tenants involved in an ongoing housing discrimination complaint. See United States v. White, 670 F.3d 498, 501, 503-04 (4th Cir.2012).

And, as the district court further observed, Appellant had also previously been convicted in the Northern District of Illinois "of soliciting the commission of a violent federal offense" against a juror "in violation of 18 U.S.C. § 373." J.A. 146 (citing United States v. White, 698 F.3d 1005 (7th Cir.2012)). In that case, Appellant authored an Internet post in which he disclosed personal information about a juror who, according to Appellant, "played a key role in convicting Matt Hale," a white supremacist. See United States v. White, 698 F.3d 1005, 1009-10 (7th Cir.2012). The posting included a picture of the juror and the juror's address and telephone numbers. See id. As the district court explained, Appellant elsewhere on the same website expressed his view that "[e]veryone associated with the Matt Hale trial has deserved assassination for a long time[.]" J.A. 146 (internal quotation marks and citation omitted).

We agree with the district court that Appellant's prior history of interfering with witnesses and a juror, and in particular his use of the Internet to publicize a juror's personal information, strongly favored the use of an anonymous jury. Appellant complains that the district court erred by affording "dispositive weight" to this factor, but the district court did not rest its decision solely on Appellant's prior history. As the district judge explained, he also found that the possibility of a lengthy sentence under § 875(b) and the considerable press attention to Appellant's trial provided additional support for the decision to empanel an anonymous jury. We perceive no abuse of discretion in that decision. Indeed, the Seventh Circuit upheld the use of an anonymous jury in a case against Appellant based on the mere allegation that he had posted the personal information of a juror on the Internet and the existence of "some publicity" around the trial. See United States v. White, 698 F.3d 1005, 1017 (7th Cir.2012) (emphasizing that Appellant was on trial for posting "the personal contact information [] of a juror").

The district court also satisfied its obligation to adopt reasonable safeguards to protect Appellant's right to a fair trial. Most significantly, the district court told the jurors they were being empaneled anonymously to prevent the press from communicating with them during trial:

J.A. 155. We have previously endorsed precisely such a safeguard, and we find it appropriate here as well. See Hager, 721 F.3d at 188-89 (observing that a very similar explanation to the jury was a sufficient "neutral non-prejudicial reason for empaneling an anonymous jury"); see also United States v. White, 698 F.3d 1005, 1017 (7th Cir.2012) (noting that any harm from empaneling an anonymous jury would be rendered harmless where the jury was told the measure was adopted "to ensure a fair and impartial trial" and the court "did not mention security as a reason"). Accordingly, the district court's use of an anonymous jury was not error.

B.

The Gnos Notes

Appellant also argues it was reversible error to admit Gnos's handwritten notes into evidence. We review evidentiary rulings for an abuse of discretion, affording substantial deference to the district court. See United States v. Medford, 661 F.3d 746, 751 (4th Cir.2011).

The relevant portions of the notes, which were admitted during Gnos's direct examination by the prosecution, read as follows:

J.A. 727-28. The Government concedes the notes are prior out-of-court statements, but argues they were admissible under Federal Rule of Evidence 803(1)'s exception for present-sense-impressions. Appellant aptly points out that the Government failed to offer that rationale for the notes' admissibility at trial. But we need not resolve the issue because, even assuming the notes are inadmissible hearsay, we agree with the Government's alternative contention that any error in admitting them was ultimately harmless.

"A nonconstitutional error ceases to be harmless if it had a substantial and injurious effect or influence in determining the jury's verdict." United States v. Briley, 770 F.3d 267, 276 (4th Cir.2014) (internal quotation marks omitted). "We do not reverse evidentiary rulings for inconsequential technicalities. Rather, reversal is reserved for more serious errors that affect substantial rights or that directly affect the outcome of a case." Id. (internal quotation marks omitted).

In this case, the notes, if believed by the jury, could have informed the jury's consideration of two important issues: whether Appellant authored the e-mails and whether Appellant intended to extort money from MW. But the Government properly introduced several audio recordings of Appellant making nearly identical comments to Gnos on several occasions. And those recordings demonstrated, in far more vibrant detail than Gnos's notes, that Appellant was preoccupied with finding someone in Virginia willing to intimidate MW.

Appellant maintains that the strength of the Government's additional evidence is not dispositive. Fair enough. But the closeness of the case, which will frequently turn on the weight of the evidence, is clearly relevant to the harmless error analysis. See Kotteakos v. United States, 328 U.S. 750, 763, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946) ("Errors of this sort in criminal causes conceivably may be altogether harmless in the face of other clear evidence, although the same error might turn scales otherwise level, as constantly appears in the application of the policy ... to questions of the admission of cumulative evidence."). And we have in the past held evidentiary errors harmless where the Government's case is strongly corroborated by other admissible evidence. See Briley, 770 F.3d at 277-78 (observing that a "plethora of testimony established" the elements of the charged offense and concluding the admission of improper character evidence was harmless).

Ultimately, the question is whether we can say "with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the [jurors'] judgment was not substantially swayed by the error." Kotteakos, 328 U.S. at 765, 66 S.Ct. 1239. In this case, because the substance of Gnos's notes was repeatedly corroborated by Appellant's own later-recorded statements, we are confident any error in admitting the notes did not affect the outcome of the case. See United States v. Mazza, 792 F.2d 1210, 1216-22 (1st Cir.1986) (Breyer, J.) (holding hearsay admission harmless where recorded conversations largely corroborated out-of-court statements).

C.

The Sufficiency of the Evidence

We next turn to Appellant's claim that he was entitled to a judgment of acquittal, a question which we review de novo. See United States v. Howard, 773 F.3d 519, 525 (4th Cir.2014). The question is whether, "viewing the evidence in the light most favorable to the prosecution, any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt." Jackson v. Virginia, 443 U.S. 307, 319, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979). Under that standard, we think a rational trier of fact could easily have found that Appellant sent MW true threats of bodily harm through foreign commerce with the intent to extort. The district court therefore properly denied the motion.

1.

Respecting the "true threat" requirement, the question — which remains unchanged following Elonis — is whether "a reasonable recipient familiar with the context" would consider the communicated statement "a serious expression of an intent to do harm." United States v. White, 670 F.3d 498, 509 (4th Cir.2012) (internal quotation marks and emphasis omitted).

In the case at hand, we have an estranged wife on the outs with her fugitive husband with whom she was engaged in a dispute over money. The e-mails comprising Counts I, III, and IV explained (either explicitly or implicitly) that the sender had been in contact with a loan shark who lived near the recipient, and advised the recipient to be ready to remit payment or risk, respectively, having someone "beat [her] ass," "probably be[ing] hospitalized," or having "something violent potentially happen[] to [her] around [her] baby." J.A. 717-25. As for the e-mail constituting Count II, it similarly warned the recipient that she would "have the living shit beat out of [her] — to start with," and that the sender had "a lot of friends who think nothing of taking out on [her] the things" she allegedly did to the sender. J.A. 721. By any measure, a reasonable person would have interpreted those messages as a serious expression of an intent to do harm. See White, 670 F.3d at 513 (holding that a caller's message that recipient would be "hunted down and shot" was a true threat).

2.

Whether Appellant sent the e-mails and whether they traveled in foreign commerce are overlapping questions. There is no dispute Appellant was in Mexico and MW in Virginia when the e-mails were sent; if he sent them, the foreign commerce element is clearly satisfied. And the jury heard more than enough evidence to find, beyond a reasonable doubt, that Appellant was indeed the author of each charged e-mail. To begin with, the e-mails originated from an e-mail address long associated with Appellant. They were also sent from an IP address that had been masked through the use of an anonymizer, technology which Appellant bragged to Gnos and on Facebook about using. Moreover, the e-mails were sent around the same time that someone using the same IP address was updating Appellant's Facebook page. Appellant flatly denied sending the e-mails, but, given the evidence just discussed, a rational trier of fact would have been justified in discounting his testimony. See United States v. Burgos, 94 F.3d 849, 862 (4th Cir.1996) (en banc) (We "remain cognizant ... that the [j]ury, not the reviewing court, weighs the credibility of the evidence and resolves any conflicts in the evidence presented." (internal quotation marks omitted)).

3.

Finally, to demonstrate that Appellant sent each e-mail with the intent to extort, the Government was obliged to show Appellant intended to induce MW to pay him by wrongfully threatening her with bodily injury if she did not. The jury heard Appellant repeatedly emphasize to Gnos the importance of finding someone to lean on MW. And each of the e-mails comprising Counts I, III, and IV specifically demands money from the recipient, and threatens violence if the money is not paid. A rational jury could easily have found that the sender of those messages intended to express exactly what those words mean — the quintessential extortionate demand: "pay up, or else." That Appellant also contemporaneously discussed his plans to intimidate MW into paying the alimony only crystalized his evident intent to extort her. Accordingly, the district court did not err in denying Appellant's motion for judgment of acquittal.5

D.

The Reasonableness of the Sentence

Finally, we consider Appellant's three challenges to his sentence. "We review the reasonableness of a sentencing decision under an abuse of discretion standard." United States v. Howard, 773 F.3d 519, 527-28 (4th Cir.2014).

1.

Appellant first asserts the district court erred in applying the two-level enhancement for obstruction of justice, rendering his sentence procedurally unreasonable. "There are three elements necessary to impose a two-level enhancement for obstruction of justice based on the defendant's perjurious testimony: the sentencing court must find that the defendant (1) gave false testimony; (2) concerning a material matter; (3) with willful intent to deceive." United States v. Perez, 661 F.3d 189, 192 (4th Cir.2011) (internal quotation marks omitted). "[I]t is preferable for a district court to address each element of the alleged perjury in a separate and clear finding. It is enough, however, if the court makes a finding of an obstruction of, or impediment to, justice that encompasses all of the factual predicates for a finding of perjury." Id. at 193 (internal quotation marks and citation omitted).

Appellant testified at trial that he did not send the threatening e-mails to MW and suggested instead that Gnos was the responsible party. The jury, inasmuch as it found him guilty, clearly rejected Appellant's alternative theory of the crime. At sentencing the district court denied Appellant's objection to the two-level enhancement. The court explained:

J.A. 1139. Although not as explicit as ideal, the district court provided a sufficient basis for imposing the two-level enhancement. The observation that the jury rejected Appellant's testimony, and the court's comment that it could "not imagine" Appellant was unaware of sending the e-mails when he testified, established falsity and willfulness. See J.A. 1139. And the issue of who authored the e-mails, which Appellant attempted to muddy by falsely implicating Gnos, was plainly material. Accordingly, the district court did not err in applying the two-level enhancement.

2.

Next, Appellant claims his sentence is substantively unreasonable because the district court improperly considered his political views. "Any sentence that is within or below a properly calculated Guidelines range is presumptively reasonable." United States v. Louthian, 756 F.3d 295, 306 (4th Cir.2014). Appellant bears the burden of convincing us that his sentence was instead greater than necessary to provide just punishment, promote respect for the law, reflect the seriousness of the offense, adequately deter similar criminal conduct, protect the public, and provide necessary rehabilitation. See 18 U.S.C. § 3553(a); see also Louthian, 756 F.3d at 306 (observing that the presumption that a within-Guidelines sentence is reasonable "can be rebutted only by showing that the sentence is unreasonable when measured against the ... § 3553(a) factors.").

Here, the district court imposed a 92-month sentence at the bottom of the Guidelines range. Appellant argues the sentence was nevertheless "greater than necessary" because it was improperly based on his unpopular political views. He relies on two statements from the bench during the sentencing hearing. In the first, the district court noted its concern that "the Government [wa]s trying to punish [Appellant] for [his] beliefs," or "largely because of [his] beliefs." J.A. 1140-41. In the second, the district court observed that it was "bothered ... a little bit" that the Guidelines range was "a little high because of [Appellant's] beliefs," but ultimately denied Appellant's request for a downward departure because his offense was "serious," and because MW "would have been very apprehensive about" receiving the threats Appellant was convicted of sending. See id. at 1145.

Appellant reads too much into these remarks. The district court immediately rebuked the Government when the prosecutor attempted to argue that some of Appellant's politically controversial writings showed a lack of respect for the law. The court explained that Appellant "has a constitutional right to believe what he believes," and reminded the Government that it "[c]an't punish him for that." J.A. 1136-37.

In sum, we think the most that can be inferred from the sentencing transcript is that the district court was concerned by the potential that Appellant had been singled out for prosecution and so selected a sentence at the bottom of the Guidelines range. But the district court also acknowledged that Appellant's words nevertheless constituted serious threats that negatively impacted MW, making departure below the Guidelines range inappropriate. We find no error in this approach.

3.

Finally, Appellant contends that the district court erred by failing to group his counts of conviction under § 3D1.2 of the Guidelines. Because he did not raise the grouping issue until his reply brief in this court, he acknowledges that plain error review is appropriate. "To satisfy plain error review, the defendant must establish that: (1) there is a sentencing error; (2) the error is plain; and (3) the error affects his substantial rights." United States v. Aplicano-Oyuela, 792 F.3d 416, 422 (4th Cir.2015). "If the three-part plain error test is satisfied, we must decide whether to cure the error, and should not do so unless the error seriously affects the fairness, integrity or public reputation of judicial proceedings." Id. (internal quotation marks and citation omitted).

Appellant's PSR stated that his convictions were not subject to grouping pursuant to § 3D1.2(d) of the Guidelines. That is a correct statement of § 3D1.2(d). But Appellant maintains it was nevertheless plain error not to group his offenses under § 3D1.2(a) or (b) and that the probation officer and district court (along with the parties' counsel) plainly misinterpreted Subsection (d) as a blanket or overriding bar against grouping under Subsection (a) or (b).

Several courts have made clear that offenses excluded from grouping under Subsection (d) may nevertheless be grouped pursuant to Subsection (a) or (b). See, e.g., United States v. Lopez-Urbina, 434 F.3d 750, 764 (5th Cir.2005); United States v. Tank, 200 F.3d 627, 632 (9th Cir.2000). But there is no evidence in the sentencing transcript suggesting the court interpreted Subsection (d) to absolutely bar grouping under any circumstances. Nor was it plainly erroneous for the district court to decline to group Appellant's offenses pursuant Subsection (a) or (b). In fact, though we need not definitively resolve the question, the Application Notes to those Subsections could be plausibly read to suggest Appellant's offenses were not subject to grouping. For example, respecting Subsection (a), Comment 3 states:

See U.S.S.G. § 3D1.2 cmt. 3. The commentary and explanations explicating Subsection (b) contain a similar distinction:

See id. § 3D1.2 cmt. 4. And the Application Notes further provide that Subsection (b) "does not authorize the grouping of offenses that cannot be considered to represent essentially one composite harm (e.g., robbery of the same victim on different occasions involves multiple, separate instances of fear and risk of harm, not one composite harm)." Id. Given that each of Appellant's threats against MW involved "separate instances of fear and risk of harm," the district court did not plainly err by categorizing "[e]ach message []as a separate offense," J.A. 1132, and declining to group them.

In support of his argument, Appellant points to United States v. Thomas, 155 F.3d 833, 840 (7th Cir.1998), in which the court remanded for consideration of the grouping question with respect to multiple § 876 convictions. But in doing so the Thomas court relied on Application Note 3 of § 2A6.1 of the Guidelines, which specifically provides that "multiple counts involving making a threatening or harassing communication to the same victim are grouped together under [sect] 3D1.2[.]" U.S.S.G. § 2A6.1 cmt. 3 (emphasis supplied). By contrast, Appellant's § 875(b) offenses fall under Guideline § 2B3.2 which says nothing about grouping; only his lone § 875(c) conviction is covered by § 2A6.1. Accordingly, given that the Application Notes to § 3D1.2 do not unambiguously direct grouping of § 875(b) offenses pursuant to Subsection (a) or (b), the district court did not plainly err by failing to do so.

IV.

A defendant is entitled to a fair trial, not a perfect one. See Delaware v. Van Arsdall, 475 U.S. 673, 681, 106 S.Ct. 1431, 89 L.Ed.2d 674 (1986). Here, Appellant received a fair trial, and we find no reason to disturb the jury's verdict or the district court's sentence. For the foregoing reasons, the judgment of the district court is

AFFIRMED.