Full opinion text
OPINION OF THE COURT FUENTES, Circuit Judge: On July 14, 2010, the United States District Court for the Eastern District of Pennsylvania sentenced former Pennsylvania State Senator Vincent J. Fumo to 55 months’ imprisonment, a $411,000 fine, and $2,340,839 in restitution, arising from his jury conviction on 137 counts of fraud, tax evasion, and obstruction of justice. A week later, the District Court sentenced former Fumo aide Ruth Arnao to imprisonment of one year and one day, a $45,000 fine, and joint and several restitution with Fumo of up to $792,802, arising from her jury conviction on 45 counts of fraud, tax evasion, and obstruction of justice. On appeal, the Government argues that the District Court made numerous procedural errors in arriving at both sentences. In particular, the Government asserts that the District Court failed to announce a final guidelines sentencing range for Fumo. Fumo cross-appeals, contending that the District Court erred when it denied his motion for a new trial based on alleged jury partiality and the District Court’s admission of evidence related to Pennsylvania’s public employee ethics law. For the following reasons, we will affirm Fumo’s conviction, vacate the sentences of Fumo and Arnao, and remand both for resentencing before the District Court. I. A. Background Vincent Fumo was a high-profile Pennsylvania state senator at the center of one of the largest political scandals in recent state history. Fumo was first elected to the State Senate in 1978 from a district in South Philadelphia. He eventually became Chairman of the Senate Democratic Appropriations Committee, which put him in control of millions of dollars that could be dispensed at his discretion for legislative purposes. Fumo served in the Pennsylvania State Senate for thirty years, where it is widely agreed that he became one of the most powerful political figures in the state. During his three decades as a state senator, Fumo frequently directed his publicly paid Senate employees to attend to his personal needs and political interests during their working hours, as well as at night and on weekends. Fumo’s Philadelphia district office was staffed by ten such employees, whose duties included providing constituent services to the residents of Fumo’s district. However, the staffers often also provided Fumo with campaign and personal assistance: organizing political fundraisers and mailings, processing bills for business accounts, and handling various aspects of Fumo’s personal finances. Various aides also acted as his housekeeper, drove him from place to place, managed the refurbishment of his 33-room house, ran personal errands, and even drove his daughter to school. During Fumo’s annual trip to Martha’s Vineyard, Massachusetts, his Senate aides would drive two vehicles from Philadelphia and back, filled with the luggage of Fumo and his guests. Staffers also used their time to assist a Philadelphia City Councilman who was Fumo’s ally and, for two months, to advance the campaign of an ultimately unsuccessful Pennsylvania Democratic gubernatorial candidate. Moreover, Fumo misused his Senate staff in Harrisburg— several of them renovated and developed a farm he had purchased in 2003 as a residential and business enterprise. In exchange, Fumo arranged salaries for his employees that were substantially greater than those designated by the State Senate for comparable Senate employees. Fumo also provided non-staffers, such as contractors, family members, and girlfriends with access to Senate resources, including laptops and computer assistance. Further, he used Senate funds to hire contractors for non-legislative tasks. For instance, Fumo obtained a $40,000 state contract for a private investigator who, in addition to his legitimate activities, conducted surveillance on Fumo’s former wife, girlfriends, ex-girlfriends’ boyfriends, and at times, political rivals. He obtained an $80,000 state contract for a consultant who spent much of his time assisting Fumo with political races and a $45,000 salary for an individual who spent most of his time assisting with Fumo’s farm. Mitchell Rubin, the boyfriend and later husband of Ruth Arnao, was paid $30,000 per year for five years, without doing much, if any, work at all. In order to facilitate his use of public funds for his own purposes, Fumo falsely represented that employees and contractors receiving payment by the Senate were performing proper and legitimate legislative functions that they only partially or never in fact completed, and failed to disclose the private and political services that they were actually performing. Fumo also provided false job descriptions and elevated position classifications that conflicted with the duties that employees actually carried out. In 1991, Fumo and his staff founded a non-profit organization that became known as the Citizens Alliance for Better Neighborhoods (“Citizens Alliance”). Arnao, a Senate employee on Fumo’s staff, became its director. Citizens Alliance’s stated purpose was to improve Philadelphia neighborhoods through projects such as removing trash, sweeping streets, trimming trees, clearing snow, and cleaning alleys and abandoned lots. Citizens Alliance received much of its funding from grants obtained by Fumo from the state and other entities. In 1998, after Fumo brought litigation challenging its utility rates, the Philadelphia Electric Company (“PECO”) privately agreed to donate $17 million to Citizens Alliance as part of a settlement agreement. The existence of the $17 million contribution only became public knowledge in November 2003, when it was reported by the Philadelphia Inquirer. After the influx of $17 million, Citizens Alliance expanded the scope of its work, acquiring properties for renovation, opening a charter school, and attempting to develop an office building for high-tech companies. However, concurrent with its expanded efforts, Fumo and Arnao began to use Citizens Alliance funds for their personal benefit, including $90,000 for tools and $6,528 for vacuum cleaners and floor machines used in Fumo’s homes. Citizens Alliance also provided Fumo and his staff with vehicles, including a $38,000 minivan, a $52,000 luxury SUV, and a $25,000 jeep. In total, more than $387,325 went towards acquiring and maintaining vehicles for the use of Fumo, Arnao, legislative aides, and family members. Further, Citizens Alliance became the landlord of Fumo’s office on Tasker Street in Philadelphia. While the Senate spent $90,000 in rent during a five-year period, Citizens Alliance spent over $600,000 to furnish, maintain, and rent Fumo’s office to him at a discount. The office also served as his campaign office and ward headquarters. Further, Citizens Alliance paid for cell phones for many of Fumo’s staffers, as well as his daughter. It also paid $39,000 for Fumo’s trip to Cuba with five friends and $50,000 for a “war dog” memorial in Bucks County- Fumo used Citizens Alliance in violation of federal 501(c)(3) rules for charitable organizations by having it pay $250,000 for political polling, $20,000 for a lawsuit against a Senate rival, and $68,000 to support opposition to the Government’s construction of dunes along the Jersey shore, which would have blocked his seaside house’s view of the ocean and reduced its property value. In order to oppose the dunes, Fumo had his Senate counsel create a nonprofit entity called “Riparian Defense Fund, Inc.” to funnel funds from Citizens Alliance, and then misled the IRS and Pennsylvania Secretary of State as to the nature and purpose of the organization. Further, Fumo misrepresented political and campaign expenses as “community development consulting” expenses on Citizens Alliance’s tax filings, deceiving the IRS yet again. Just as he had done with his public employees, Fumo directed Citizens Alliance staff to assist with his personal matters, traveling to his house on the Jersey shore to repair and paint his dock and deck, picking up trash, and undertaking other errands and tasks. They also frequently cleaned and served his Philadelphia home, and delivered equipment and personal items to his farm. Additionally, Citizens Alliance paid for a $27,000 bulldozer, a lawn tractor, a dump truck, an all-terrain vehicle, and a Ford F-150 pickup truck for his Harrisburg-area farm. Fumo and Arnao never disclosed the funds used for Fumo’s personal benefit to Citizens Alliance’s accountants, and when asked about those funds by an accountant, Arnao misstated their purpose. Fumo and Arnao also made repeated misrepresentations to journalists about .Citizens Alliance and how it spent its funds. Fumo served on the board of directors of the Independence Seaport Museum (“ISM”). Board members did not receive compensation or benefits from the museum, but were expected to help the museum develop and solicit donors. While Fumo did not donate or solicit much in the way of donations for the ISM, he did use his influence to obtain grants for the museum from the state and other entities. However, at the expense of the ISM, he also repeatedly used its yachts for pleasure cruises and its ship models for decorations in his home and office. These personal uses of the ISM’s resources, which were approved by ISM’s president John Carter, were in violation of the museum’s policies and bylaws. Fumo later claimed that he used the yachts to help raise money for the museum and that he sometimes paid for their use. In 2003, the Government began investigating Fumo. In December, the Philadelphia Inquirer published a series of articles about Citizens Alliance’s use of funds and its relationship with Fumo. Shortly thereafter, Fumo directed a computer technician on his staff to ensure that all emails to and from Fumo and others were deleted. When the Inquirer ran an article entitled “FBI Probes Fumo Deal” on January 25, 2004, Fumo involved additional Senate aides and expanded the scope of his attempts to delete emails. Throughout 2004 his aides, including Arnao, deleted email from numerous computers and communication devices, and then “wiped” the computers using sophisticated programs in order to prevent forensic analysis. These efforts included wiping computers at Arnao’s home and at Citizens Alliance. Despite Fumo’s efforts, two of the aides involved in the deletion kept emails between each other, including emails regarding Fumo’s instructions to eliminate computer evidence of the fraud. B. The Trial The Government charged Fumo and Arnao under what was to later become a 141 count superseding indictment. Counts 1 through 64 related to fraud on the Pennsylvania State Senate, Counts 65 through 98 to fraud on Citizens Alliance, Counts 99 through 103 to tax evasion by Citizens Alliance, Counts 104 through 108 to fraud on ISM, and Counts 109 through 141 to obstruction of justice and conspiracy to commit obstruction of justice. Fumo was charged in 139 counts, including all but Counts 100 and 102. At trial, the Government voluntarily moved to dismiss Counts 36 and 38 against Fumo. Arnao was charged in 45 counts, including Counts 65 through 98, related to the fraud on Citizens Alliance, Counts 99, 100, and 102, related to tax evasion, and Counts 109, 121, 124, 126, 127, 129, 132, and 134, related to obstruction of justice. The case was originally assigned to the Honorable William H. Yohn, Jr., and after some delay while Fumo found satisfactory defense counsel, jury selection began on September 8, 2008. After the case was reassigned to the Honorable Ronald L. Buckwalter, jury selection resumed on October 20, 2008. The trial lasted an additional five months, with the proceedings halted on Fridays. By the time it rested its case on January 26, 2009, the Government had called 80 witnesses in its case-in-chief. The defendants then called an additional 25 witnesses, including Fumo himself, and rested their case on February 18, 2009. On March 16, 2009, after four days of deliberation, the jury convicted Fumo of all 137 counts presented against him, and Arnao of all 45 counts presented against her. A number of events occurred during the trial that Fumo now asserts as the bases for his cross-appeal. First, during the trial, the Government called John J. Contino as an expert witness to testify about the Pennsylvania Public Official and Employee Ethics Act, 65 Pa. Con. Stat. Ann. § 1101, et seq. (the “Ethics Act”). Contino is the Executive Director of the State Ethics Commission (the “Commission”), the body charged with enforcing the Ethics Act. Section 1103(a) of The Ethics Act prohibits a public official or employee from engaging in conduct that constitutes a “conflict of interest,” which is defined at § 1102 as the “[u]se by a public official or public employee of the authority of his office or employment ... for the private pecuniary benefit of himself, a member of his immediate family or a business with which he or a member of his immediate family is associated.” Prior to the trial, Judge Yohn had found Contino to be “well qualified” as an expert and ruled that it was “appropriate for him to talk about the Ethics Act.” (J.A. 431). During trial, Contino testified as to how and to whom the Ethics Act applied, whether it was mandatory in nature, and as to how the legislature was apprised of the Ethics Act and the Commission’s interpretation of it. Contino also referenced abridged versions of the Commission’s opinions, summarizing violations that were considered and ruled upon by the Commission. He did not, however, express an opinion as to whether Fumo’s own actions violated the Ethics Act or whether Fumo was guilty of the federal charges against him. The Government also extensively cross-examined Fumo on the subject of the Ethics Act and specifically his knowledge and understanding of it. At the time of the cross-examination, the District Court provided a limiting instruction to the jury, reminding them that no law required Fumo to study the decisions or reports of the Commission. At the conclusion of the trial, the District Court further instructed the jury on the Ethics Act, telling them that they could “consider [such] evidence ... to the extent that [they] find it sheds light on questions of willfulness, intent to defraud, and good faith” but that “violation of the ethics laws should not be considered by [them] as implying a violation of federal criminal law” and that they “may not convict Fumo of any of the counts alleging that he conspired or attempted to execute a scheme to defraud the Senate of money or property simply on the basis of the conclusion that he may have violated a state ethics law.” (J.A. 4363). On March 15, 2009, while jury deliberations were ongoing, a local television station reported that one of the jurors, hereinafter referred to as “Juror 1,” had made postings on both his Facebook and Twitter pages related to the trial. That night, which was the night before the jury returned its verdict, Juror 1 was watching television when he learned that the media was following the comments he had made on the internet. He subsequently panicked and deleted the comments from his Facebook page. Prior to deleting them, Juror 1 made the following comments on his Facebook “wall” during jury selection and the trial: — Sept. 18: (apparently upon a continuance of the trial due to judge’s illness): “[Juror 1] is glad he got a 5 week reprieve, but still could use the money — Jan. 11: (apparently referring to the end of the government’s case): “[Juror 1] is wondering if this could be the week to end Part 1?” — Jan. 21: “[Juror 1] wonders if today will really be the end of Part 1???” — Mar. 4: (conclusion of closing arguments): “[Juror 1] can’t believe tomorrow may actually be the end!!!” — Mar. 8: (Sunday evening before second day of deliberations): “[Juror 1] is not sure about tomorrow ...” — Mar. 9: (end of second day of deliberations): “[Juror 1] says today was much better than expected and tomorrow looks promising too!” — Mar. 13: (Friday after completion of first week of deliberations): “Stay tuned for the big announcement on Monday everyone!” (J.A. 587-88). Juror l’s Facebook comments appeared over the many months of the trial, and in the midst of dozens of other comments he made unrelated to the trial. It was the final, March 13 post that was the subject of media attention. With regard to Twitter, Juror 1 made a single comment or “tweet” on March 13, stating “This is it ... no looking back now!” (J.A. 587). When Fumo learned of Juror l’s Face-book and Twitter comments, he moved to disqualify Juror 1 from the jury. The District Court held an in camera review of the issue, and questioned Juror 1 about his activities on these two websites and his general media consumption. Juror 1 told the judge that he saw the news report that night because he had been watching another show when the local news began. He nevertheless explained that he had avoided television news during the entire trial. He also affirmed that he had not discussed the substance of the case with anyone. Juror 1 further stated that he had made the comments “for my benefit to just get it out of my head, similar to a blog posting or somebody journaling something.” (J.A. 589). In a written opinion, the District Court determined that there was no evidence that Juror 1 received outside influence due to his Facebook or Twitter postings and concluded that, although in violation of his instruction not to discuss the case outside of the jury room, they were “nothing more than harmless ramblings having no prejudicial effect. They were so vague as to be virtually meaningless.” (J.A. 592). More than three months after the verdict, but before sentencing, Fumo filed a second motion for a new trial, attaching the affidavit of counsel Dennis Cogan. The affidavit asserted that journalist Ralph Cipriano, writing for Philadelphia Magazine, had contacted Cogan regarding information he obtained during post-verdict interviews with several jurors. According to an article written by Cipriano, on the morning of March 16, the day of the verdict, all of the jurors had heard media reports about Juror l’s use of Facebook and Twitter. Further, another juror hereinafter referred to as “Juror 2,” indicated that while at her workplace on a Friday, several co-workers informed her of Fumo’s prior overturned conviction, as well as the conviction and imprisonment of John Carter, former president of the ISM. Both of these facts had previously been excluded from the trial by the District Court. Specifically, the article stated that Juror 2 had told Cipriano that: Co-workers stopped by and talked about things in the media, such as Fumo’s prior 1980 conviction, subsequently overturned by a judge, for hiring ghost employees. Judgé Buckwalter repeatedly turned down prosecution requests to tell the jury about that prior conviction. But [Juror 2] found out anyway, even though she held up her hands and told co-workers: Please don’t talk to me, I can’t discuss the case. Co-workers also told her that John Carter, former president of the Independence Seaport Museum, and the guy who gave Fumo permission to take free yacht trips, was doing time for fraud. The judge didn’t want the jury to know about Carter, either. (J.A. 703-04) (emphasis in original). There was no evidence that any other juror had learned of Fumo’s prior conviction or the conviction of Carter, and the other five jurors interviewed by Cipriano did not mention either fact. The District Court denied the motion, concluding that the information was an insufficient basis to hold a hearing and that, even if everything asserted by Juror 2 were true, it would not constitute the showing of substantial prejudice required to grant a new trial. C. Sentencing On July 8, the District Court held a sentencing hearing at which the parties made arguments directed at the sentencing guidelines calculations for both Fumo and Arnao. The Government adopted the position of the Pre-sentence Report (“PSR”), which divided Fumo’s crimes into two groups pursuant to § 3D1.2 of the Sentencing Guidelines — the first made up of the 134 fraud and obstruction of justice counts, and the second consisting of the three tax evasion counts (Counts 99, 101, and 103). As to the first group, the PSR began with a base offense level of 7 under U.S.S.G. § 2Bl.l(a)(l). It then added 18 levels under § 2Bl.l(b)(l)(J) because it calculated the loss from the fraud to be greater than $2,500,000, and specifically $4,339,041. The PSR then added 2 levels under § 2Bl.l(b)(8)(A) because it concluded Fumo misrepresented that he was acting on behalf of a charitable organization, Citizens Alliance. Similarly, it added 2 levels under § 2Bl.l(b)(9)(C) because the fraud involved the use of sophisticated means, in that Fumo used a shell corporation, Eastern Leasing Corp., to purchase vehicles for his personal use and conduct political polling, and used a consulting firm as a conduit to conceal his role in a lawsuit against one of his political rivals. The PSR added an additional 4 levels under § 3Bl.l(a) for Fumo’s role as the organizer or leader of the fraud, and 2 levels under § 3B1.3 because he was in a position of public trust. Finally, under § 3C1.1, it added 2 levels for Fumo’s obstruction of justice during the investigation of the offense, and 2 levels for his obstruction of justice in perjuring himself at trial. In total, the PSR calculated Fumo’s adjusted offense level for the fraud group as 39. As to the tax evasion group, the PSR began with a base offense level of 24 under §§ 2Tl.l(a)(l) and 2T4.1(J) because the tax loss was more than $2,500,000, and specifically $4,624,300. It then added 2 levels under § 2Tl.l(b)(2) because the offense involved sophisticated means, for a total adjusted offense level of 26. Because the tax evasion group’s offense level of 26 was more than 8 levels below the fraud group’s offense level of 39, pursuant to § 3D1.4(c), no additional levels were added to the larger of the two. Accordingly, the PSR calculated, and the Government argued, that the District Court should find Fumo’s total adjusted offense level to be 39 and his criminal history category to be I, which would mean a guideline range of 262 to 327 months’ imprisonment. The day after the July 8 hearing, the District Court issued an order ruling that it would not apply the 2-level enhancement for charitable misrepresentation, the 2-level enhancement for sophisticated means, or the second 2-level obstruction of justice enhancement for perjury at trial. It also calculated the total loss from the fraud to be $2,379,914 — about $2,000,000 less than the Government’s calculation and a reduction of 2 additional levels. The District Court also declined to apply the 2-level enhancement for sophisticated means to the tax evasion group. Additionally, Fumo requested two downward departures based on his physical health under § 5H1.4 and for extraordinary public service under § 5H1.11. The District Court denied the former and reserved judgment on the latter until the final sentencing hearing. With reduced adjusted offense levels of 31 and 24 for the fraud and tax evasion groups, respectively, the combined offense level became 32 under § 3D1.4(b), translating into a guideline range of 121 to 151 months’ imprisonment. On July 14, the District Court held another lengthy hearing. When the Government learned that the Court had calculated a guideline range of 121 to 151 months, it sought an upward variance, arguing that the adjusted range did not adequately represent or take into account the full loss from the fraud, the damage to public institutions, Fumo’s alleged perjury at trial, other obstructive conduct, and Fumo’s alleged lack of remorse. The District Court declined to vary upwards. It also denied Fumo’s request for a departure on the basis of his medical condition. Then, after hearing from six witnesses who spoke on Fumo’s behalf, and reviewing hundreds of letters from the public, it found that Fumo had “worked hard for the public and ... worked extraordinarily hard” such that it would “grant a departure from the guidelines.” (J.A. 1622-23). Without enunciating any modification to the guideline range of 121 to 151 months, the District Court then sentenced Fumo to a term of imprisonment of 55 months, three years of supervised release, a $411,000 fine, a $13,700 special assessment, $2,084,979 in restitution, and $255,860 in prejudgment interest on the restitution. Fumo filed a Motion for Correction of Sentence under Federal Rule of Criminal Procedure 35(a), asking the Court to resolve various issues related to the sentence. Among the issues raised was the fact that the District Court had, during the July 14 sentencing hearing, three times referred to the sentence as a “departure” from the guidelines range. The motion papers noted that “[w]hen a sentencing court grants a true ‘departure,’ [as opposed to a variance,] it must ‘state how the departure affects the Guidelines calculation.’ This Court[ ] fail[ed] to make such a statement....” (J.A. 1629) (quoting United States v. Tomko, 562 F.3d 558, 567 (3d Cir.2009) (en banc)). They also suggested that “[i]n context, it appears that the Court intended the sentence as a statute-based ‘variance,’ designed to achieve a punishment sufficient but not greater than necessary to fulfill the objectives set forth at 18 U.S.C. § 3553(a)(2), rather than as a Guidelines Manual-based ‘departure.’ ” (J.A. 1629). Fumo asked that the Court “correct this technical error.” (J.A. 1629). The Government filed a response, contesting Fumo’s characterization of the Court’s below-guideline sentence as a variance and noting that “the Court repeatedly stated that it decided to grant the departure motion based on public service.” (J.A. 1635). The following day, the District Court issued a Memorandum and Order, which among other things, explained that “[t]he government correctly states that the court announced it was granting a departure. Thereafter, the court never enunciated the guideline level to which it departed, and, in fact, never reached the sentence it did by consulting any specific level on the guideline chart.” (J.A. 1653). The District Court also filed a Judgment and a formal Statement of Reasons. The Statement read, in pertinent part: I next determined whether there should be a departure from the guidelines and announced at the sentencing hearing that there should be based on my finding extraordinary good works by the defendant. I did not announce what specific guideline level the offense fell into; that is to say, the precise number of levels by which I intended to depart because until I considered all other sentencing factors, I could not determine in precise months the extent that I would vary from the guidelines. Having advised counsel of the offense level that I found and my intent to depart downward, I then proceeded to hear from counsel their respective analyses of what an appropriate sentence should be. The procedure I followed was perhaps more akin to that associated with a variance than a downward departure because I never announced nor have I ever determined to what guideline level I had departed. Ultimately, the argument over which it was elevates form over substance. (Sealed App. 185-86). The Statement of Reasons further indicated that the Court had granted Fumo a departure under § 5H1.11 of the Sentencing Guidelines for “Military Record, Charitable Service, Good Works.” After sentencing Fumo, the District Court held a sentencing hearing for Arnao. The PSR originally recommended, and the Government argued, that the loss from Arnao’s fraud was between $1 and $2.5 million, leading to an offense level of 23 under § 2Bl.l(b)(l)(I) of the Sentencing Guidelines. The PSR also recommended 2-level enhancements for the use of sophisticated means, misrepresentation on behalf of a charitable organization, and obstructions of justice, generating a total adjusted offense level of 29. Just as for Fumo, the PSR’s offense level calculation for the tax evasion group began with a base offense level of 24 and then added 2 levels because the offense involved sophisticated means, for a total adjusted offense level of 26. Under the grouping rules of § 3D1.4, two additional levels were added to the higher offense level of 29, making the combined offense level 31. With a criminal history category of I, this entailed a sentencing range of 108 to 135 months. At the hearing, the District Court rejected the sophisticated means enhancement and determined that the loss from the Citizens Alliance fraud was less than $1,000,000, and specifically $958,080, thus reducing the fraud and tax evasion group offense levels to 25 and 24, respectively. This created a combined total offense level of 27 under the grouping rules of § 3D 1.4 and a guidelines sentencing range of 70 to 87 months. The District Court then imposed a sentence of one year and one day — a substantial downward variance — to run concurrently on all counts, three years’ supervised release, a $45,000 fine, a $4,500 special assessment, and restitution to Citizens Alliance in the amount of $792,802, jointly and severally with Fumo. II. Appeal of Fumo’s conviction A. Evidence relating to the Pennsylvania Ethics Act In his appeal of the conviction, Fumo argues that the evidence presented by the Government with regard to the state Ethics Act was irrelevant to the federal criminal charges against him, and was highly prejudicial because it was likely to confuse the jury and suggest that Fumo was in violation of state law. The District Court’s rulings regarding the admissibility of evidence and expert testimony are reviewed for abuse of discretion. United States v. Mathis, 264 F.3d 321, 335 (3d Cir.2001); United States v. Serafini, 233 F.3d 758, 768 n. 14 (3d Cir.2000). The Government responds that evidence regarding the Ethics Act was of substantial relevance because it was necessary to show that the Senate did not approve of the kind of expenditures Fumo made using state money, as well as to show that Fumo intended to deceive the Senate by misleading it about how he was spending that money. The Government notes that this was particularly true given Fumo’s initial theory of the case at trial — that no rules or laws barred employing Senate resources for his personal use, or that if there were such rules, that they were entirely vague, unclear, and unenforced. Fumo also initially planned to call three experts regarding their experiences with the “customs and practices of the Senate,” focusing specific attention on “accepted uses of staff and other resources as they comport with the Ethics Act.” (Gov.Supp.App.64). In light of Fumo’s theory of the case, the content and enforcement of the Ethics Act was clearly relevant to the Government’s claim that there were rules that Fumo broke repeatedly, that those rules were clear enough for him to understand, and to show that he was deceiving the Senate when he misrepresented or omitted aspects of his actions and expenditures to avoid the perception that he had violated those rules. Without this evidence, it would have been very difficult for the Government to prove fraudulent intent. See United States v. Copple, 24 F.3d 535, 545 (3d Cir.1994) (“Proving specific intent in mail fraud cases is difficult, and, as a result, a liberal policy has developed to allow the government to introduce evidence that even peripherally bears on the question of intent.”). Further, the District Court read the jury a jointly drafted instruction, both during the trial and after the closings, which emphasized that Fumo was not on trial for violating the Ethics Act, and that even a violation of the Ethics Act by itself did not imply that he defrauded or conspired to defraud the Senate. The District Court’s finding that evidence related to the Ethics Act was relevant and not unfairly prejudicial was not an abuse of discretion. Similarly, it was not an abuse of discretion for the District Court to permit John Contino, the Director of the State Ethics Commission, to testify about the Ethics Act. We have previously explained that “[wjhile it is not permissible for a witness to testify as to the governing law since it is the district court’s duty to explain the law to the jury, our Court has allowed expert testimony concerning business customs and practices.” United States v. Leo, 941 F.2d 181, 196 (3d Cir. 1991). These customs and practices will sometimes include applicable legal regulations, such as registration requirements for securities registration under the Securities Acts, Berckeley Inv. Grp., Ltd. v. Colkitt, 455 F.3d 195, 218-19 (3d Cir.2006), or Medicaid rules, United States v. Davis, 471 F.3d 783, 789 (7th Cir.2006). Similarly, expert testimony may also concern ethics rules and laws related to public officials and government contractors. Appropriately, Contino never testified as to whether Fumo himself had violated the Ethics Act, or whether he was guilty of any of the crimes with which he was charged. Contino also properly explained the Commission’s disciplinary proceedings, its advisory opinions, and the annual report it publishes, which is distributed to every state legislator. This was evidence relevant to the question of whether Fumo was aware of the Senate ethics rules, and thus had an intent to defraud when he represented and omitted facts in a way that made him falsely appear to be in compliance with those rules. Part of Contino’s explanation of the seriousness and mandatory nature of the rules was a description of some of the Commission’s disciplinary opinions, and the penalties that were imposed for violations of the rules. The Government also properly posed questions to Contino about whether certain hypothetical facts would constitute violations of the Ethics Act — a line of questioning it had suggested in its pretrial disclosures and later pursued in light of Fumo’s theory of the case. Finally, the Government’s cross-examination of Fumo on the subject of the Ethics Act was also appropriate. During direct examination, Fumo testified that “there are no rules[,]” as to his exercise of discretion regarding spending and that “there are no guidelines” as to whether staffers can do personal errands for lawmakers. (J.A. 3967). He then claimed that “none of this is written down anywhere, and I think it’s left up to the discretion of the senator to do that as you see fit and appropriate and as you need it.” (J.A. 3967). Accordingly, in order to impeach this testimony, the Government understandably questioned Fumo about his familiarity with the annual reports of the Commission that were sent to him personally. Fumo denied ever having read the annual reports of the Commission, although he admitted being aware of them. Yet merely because this line of questioning did not turn out to be directly fruitful for the Government — although it very well may have undermined Fumo’s credibility-does not mean that it was irrelevant or unfairly prejudicial. As a precaution, however, the District Court instructed the jury that Fumo was, among other things, not required to have read the annual reports. In sum, the District Court was well within the bounds of its discretion in admitting the expert testimony of Confino and permitting the cross-examination of Fumo on the issue of the Ethics Act. B. Challenges to the jury’s fairness and impartiality Fumo challenges two rulings of the District Court denying his motions for a new trial on account of jurors’ exposure to extraneous information, and the purported prejudice and partiality that may have resulted. We review a court’s order “which denies a new trial based on alleged prejudicial information for abuse of discretion.” United States v. Urban, 404 F.3d 754, 777 (3d Cir.2005) (internal quotation and citation omitted). “A new trial is warranted if the defendant likely suffered ‘substantial prejudice’ as a result of the jury’s exposure to the extraneous information.” Id. (quoting United States v. Lloyd, 269 F.3d 228, 238 (3d Cir.2001)). “In examining for prejudice, we must conduct an objective analysis by considering the probable effect of the allegedly prejudicial information on a hypothetical average juror.” Id. (quoting Lloyd, 269 F.3d at 238 (internal quotation omitted)). Yet, the “court may inquire only into the existence of extraneous information” and not “into the subjective effect of such information on the particular jurors.” Wilson v. Vermont Castings Inc., 170 F.3d 391, 394 (3d Cir. 1999). “If there is reason to believe that jurors have been exposed to prejudicial information, the trial judge is obliged to investigate the effect of that exposure on the outcome of the trial.” United States v. Console, 13 F.3d 641, 669 (3d Cir.1993) (internal quotation omitted). However, the court is not required to conduct an investigation where an insufficient factual basis for it exists. Id. Further, even if a foundation has been established for the claim, the court need not hold a hearing “at the behest of a party whose allegations if established would not entitle it to relief.” United States v. Gilsenan, 949 F.2d 90, 97 (3d Cir.1991). Accordingly, if the Court declines to hold a hearing, it must assume that the party seeking the hearing is able to prove that the jury was presented with extraneous information, id., and determine whether “the defendant likely suffered ‘substantial prejudice’ as a result of the jury’s exposure.” Lloyd, 269 F.3d at 238 (internal citation omitted). 1. Juror l’s comments on Facebook and Twitter Fumo argues that Juror l’s comments on Facebook and Twitter brought widespread public attention to the jury’s deliberations, creating a “cloud of intense and widespread media coverage ... and [the] public expectation that a verdict [wa]s imminent[,]” thereby violating his Sixth Amendment right to a fair and impartial trial. (Cross-App’t Br. 131). Fumo also argues that the fact that Juror 1 watched the evening news, in which his own internet comments were discussed, implies or suggests that he may have been compromised by bias or partiality. In 2009, the Judicial Conference Committee on Court Administration and Case Management published proposed model jury instructions regarding “The Use of Electronic Technology to Conduct Research on or Communicate about a Case.” While the instructions focus on the importance of jurors not consulting websites or blogs to research or obtain information about the case, they also caution and instruct jurors on the use of social media: Before Trial: Until you retire to deliberate, you may not discuss this case with anyone, even your fellow jurors. After you retire to deliberate, you may begin discussing the case with your fellow jurors, but you cannot discuss the case with anyone else until you have returned a verdict and the case is at an end. I hope that for all of you this case is interesting and noteworthy. I know that many of you use cell phones, Blackberries, the internet and other tools of technology. You also must not talk to anyone about this case or use these tools to communicate electronically with anyone about the case. This includes your family and friends. You may not communicate with anyone about the case on your cell phone, through e-mail, Blackberry, iPhone, text messaging, or on Twitter, through any blog or website, through any internet chat room, or by way of any other social networking websites, including Facebook, My Space, Linkedln, and YouTube. At the Close of the Case: During your deliberations, you must not communicate with or provide any information to anyone by any means about this case. You may not use any electronic device or media, such as a telephone, cell phone, smart phone, iPhone, Blackberry or computer; the internet, any internet service, or any text or instant messaging service; or any internet chat room, blog, or website such as Facebook, My Space, Linkedln, YouTube or Twitter, to communicate to anyone any information about this ease or to conduct any research about this case until I accept your verdict. Proposed Model Jury Instructions: The Use of Electronic Technology to Conduct Research on or Communicate about a Case, Judicial Conference Committee on Court Administration and Case Management, December 2009, available at http:// www.uscourts.gov/uscourts/News/2010/ docs/DIR10-018-Attachment.pdf (last visited August 22, 2011). We enthusiastically endorse these instructions and strongly encourage district courts to routinely incorporate them or similar language into their own instructions. Not unlike a juror who speaks with friends or family members about a trial before the verdict is returned, a juror who comments about a case on the internet or social media may engender responses that include extraneous information about the case, or attempts to exercise persuasion and influence. If anything, the risk of such prejudicial communication may be greater when a juror comments on a blog or social media website than when she has a discussion about the case in person, given that the universe of individuals who are able to see and respond to a comment on Facebook or a blog is significantly larger. Yet while prohibiting and admonishing jurors from commenting — even obliquely — about a trial on social networking websites and other internet mediums is the preferred and highly recommended practice, it does not follow that every failure of a juror to abide by that prohibition will result in a new trial. Rather, as with other claims of juror partiality and exposure to extraneous information, courts must look to determine if the defendant was substantially prejudiced. Here, with regard to Juror l’s posts, none of Fumo’s theories of bias or partiality is plausible, let alone sufficient for us to find that the District Court abused its discretion in denying his motion for a new trial. The District Court questioned Juror 1 in camera at length about both his comments online and his efforts to avoid media coverage of the case. The Court found no evidence that Juror 1 had been contacted regarding the posts, or that Juror 1 had been accessing media sources beyond the single incident when he accidently learned of the attention that the media and public were paying to his comments. The Court also concluded that the posts on Facebook were “so opaque that there was no possible way that members of [Facebook’s] Philadelphia network could read them and have any obvious understanding of his discussion.” (J.A. 591). It then described the posts as “nothing more than harmless ramblings having no prejudicial effect. They were so vague as to be virtually meaningless. [Juror 1] raised no specific facts dealing with the trial, and nothing in these comments indicated any disposition toward anyone involved in the suit.” (J.A. 592). We largely agree with these characterizations of the comments. Finally, the District Court found that despite violating its prohibition against discussing the details of the trial, “[Juror 1] was a trustworthy juror who was very conscientious of his duties. There was no evidence presented by either party showing that his extra-jury misconduct had a prejudicial impact on the Defendants.” (J.A. 597-98). In light of these findings, which were based in large part on Juror l’s in-person testimony and demeanor, there is simply no plausible theory for how Fumo suffered any prejudice, let alone substantial prejudice, from Juror l’s Facebook and Twitter comments. Nor does Fumo provide a plausible theory for how the fact that other jurors may have learned of Juror l’s “vague” and “virtually meaningless” comments on Facebook could have led to substantial prejudice against him. Accordingly, the District Court did not abuse its discretion when it denied Fumo’s motion for a new trial on this basis. 2. Juror 2’s exposure to excluded evidence Three months after his conviction, Fumo’s counsel alleged that Juror 2 had learned from co-workers, during the trial, about both Fumo’s prior overturned conviction for hiring ghost employees, as well as the conviction of the former ISM president, John Carter, on charges of fraud. Both of these pieces of evidence had been excluded from the trial by the District Court. In contrast to allegations of bias made during a trial, we “are always reluctant to haul jurors in after they have reached a verdict in order to probe for potential instances of bias, misconduct or extraneous influences. As we have said before, post-verdict inquiries may lead to evil consequences: subjecting juries to harassment, inhibiting juryroom deliberation, burdening courts with meritless applications, increasing temptation for jury tampering and creating uncertainty in jury verdicts.” Gilsenan, 949 F.2d at 97 (quoting United States v. Ianniello, 866 F.2d 540, 543 (2d Cir.1989)). “It is qualitatively a different thing to conduct a voir dire during an ongoing proceeding at which the jury is part of the adjudicative process than to recall a jury months or years later for that purpose.” Id. at 98. Here, the District Court rejected the foundational basis of the allegations that Juror 2 had learned of excluded evidence from co-workers. It characterized defense counsel’s double-hearsay affidavit, which recounted the reporter’s interviews with the jurors, as lacking the “clear, strong, substantial, and incontrovertible evidence that a specific, nonspeculative impropriety occurred.” (J.A. 692). We need not address the question of whether there was sufficient foundational basis for a hearing, however, because we agree with the District Court that even if everything reported by Cipriano about what Juror 2 learned from her co-workers were true, it would not be sufficient for a showing of “substantial prejudice.” We also need not determine which party has the burden of persuasion in deciding this issue, as even if the burden were on the Government to show the lack of substantial prejudice, we find that it pointed to sufficient evidence in the record for the District Court to conclude that it made such a showing. The factors we have looked to in determining whether there was substantial prejudice include whether (1) “the extraneous information ... relatefs] to one of the elements of the case that was decided against the party moving for a new trial,” Lloyd, 269 F.3d at 239; (2) “the extent of the jury’s exposure to the extraneous information; [(3)] the time at which the jury receives the extraneous information; [ (4) ] the length of the jury’s deliberations and the structure of the verdict; [ (5) ] the existence of instructions from the court that the jury should consider only evidence developed in the case[,]” Urban, 404 F.3d at 778 (quoting Lloyd, 269 F.3d at 240-41); and (6) whether there is “a heavy volume of incriminating evidenee[.]” Lloyd, 269 F.3d at 241 (internal quotation omitted). Here, while the fourth and to some extent the first factor weigh in Fumo’s favor, they are easily overwhelmed by the second, fifth, and sixth factors, which weigh heavily against a finding of substantial prejudice. First, while knowledge of Fumo’s earlier conviction had some potential for prejudice, the fact that the conviction occurred nearly thirty years prior, in 1980, as well as the fact that it was overturned, are mitigating factors. Perhaps most importantly, the fact that only one juror was exposed to a brief verbal summary of the excluded evidence from her coworkers is a compelling consideration against a finding of prejudice. See Urban, 404 F.3d at 778 (finding that the extent of the jury’s exposure to a news article “was limited to non-existent, thus supporting the absence of prejudice” where only one juror had read the prejudicial article, and four others had “looked at the picture on the first page ... or glanced at [its] contents”). Moreover, the District Court gave careful and repeated instructions to the jurors, including immediately before deliberation, that they should “not let rumors, suspicions, or anything else that [they] may have seen or heard outside of the court influence [their] decision in any way.” (J.A. 4631). Curative instructions cannot fix every mistake, but we do generally presume that juries follow their instructions. United States v. Liburd, 607 F.3d 339, 344 (3d Cir.2010). Finally, the sixth factor — the heavy volume of incriminating evidence — also weighs heavily against a finding of prejudice. The Government’s case was presented over the course of three months and included an astonishing 80 witnesses. Further, as the Government accurately explains in footnote 16 of its opening brief, “Fumo testified at trial [and] admitted many of the acts alleged in the indictment, but asserted they were not criminal----” (Appellant Br. 44 n. 16). While many of the physical facts related to the fraud were therefore largely undisputed, the active destruction of computer records related to the fraud provided particularly potent evidence of Fumo’s motive, knowledge and intent. In light of these factors, and even assuming that the Government had the burden of persuasion, the District Court did not abuse its discretion when it found that Juror 2’s exposure to extraneous information was unlikely to have led to substantial prejudice. III. Appeal of Fumo’s sentence “In sentencing a defendant, district courts follow a three-step process: At step one, the court calculates the applicable Guidelines range, which includes the application of any sentencing enhancements.” United States v. Wright, 642 F.3d 148, 152 (3d Cir.2011) (citing Tomko, 562 F.3d at 567; United States v. Shedrick, 493 F.3d 292, 298 n. 5 (3d Cir.2007)). “At step two, the court considers any motions for departure and, if granted, states how the departure affects the Guidelines calculation.” Id. (citing Tomko, 562 F.3d at 567). “At step three, the court considers the recommended Guidelines range together with the statutory factors listed in 18 U.S.C. § 3553(a) and determines the appropriate sentence, which may vary upward or downward from the range suggested by the Guidelines.” Id. (citing Tomko, 562 F.3d at 567). “Our review of a criminal sentence ... proceeds in two stages. First, we review for procedural error at any sentencing step, including, for example, failing to make a correct computation of the Guidelines range at step one, failing to rely on appropriate bases for departure at step two, or failing to give meaningful consideration to the § 3553(a) factors at step three.” Id. (internal citations and quotations omitted). “If there is no procedural error, the second stage of our review is for substantive unreasonableness, and we will affirm the sentence unless no reasonable sentencing court would have imposed the same sentence on that particular defendant for the reasons the district court provided.” Id. (quoting Tomko, 562 F.3d at 568) (internal quotation omitted). Here, the Government does not challenge the substantive reasonableness of either Fumo’s or Arnao’s sentence — it only alleges procedural error. “The abuse-of-discretion standard applies to both our procedural and substantive reasonableness inquiries.” Tomko, 562 F.3d at 567 (citing Gall v. United States, 552 U.S. 38, 51, 128 S.Ct. 586, 169 L.Ed.2d 445 (2007); United States v. Wise, 515 F.3d 207, 217-18 (3d Cir.2008)). “For example, an abuse of discretion has occurred if a district court based its decision on a clearly erroneous factual conclusion or an erroneous legal conclusion.” Id. at 567-68 (citing Wise, 515 F.3d at 217). Our dissenting colleague argues that the proper standard of review for the District Court’s failure to arrive at a final guideline range is plain error because the Government did not object to this failure in its sentencing memoranda or at the sentencing hearing. (Dissenting Op. at 324-26). However, at the July 8 sentencing hearing the Government argued the merits of and objected to Fumo’s proposed departures. It also made its position clear that the District Court must first “determine whether there are grounds for departure and, if so, how many levels up or down ... thus reaching a final guideline range ” before “then ... applying] all of the 3553(a) factors, one of which, of course, is the guideline range that [the Court calculated].” (J.A. 1558) (emphasis added). In light of these arguments, and the District Court’s failure to advise the parties that it would not separately calculate a final guideline range after the completion of step two, the Government could not have foreseen that the District Court would fail to determine the extent of the departure when it pronounced its sentence. As our colleague notes, “the Government could not have objected because the decision it claims on appeal to be error had not even been made.” (Dissenting Op. at 326). Under these circumstances, including the lack of an opportunity to object to the District Court’s procedures prior to its pronouncement of sentence, we conclude that the Government’s substantive objections to Fumo’s departure requests as well as its recitation, to the Court, of the three-step sentencing process preserve its claim for appellate review. See United States v. Sevilla, 541 F.3d 226, 230-31 (3d Cir.2008) (defendant’s failure to object “at close of sentencing” to the district court’s neglect of sentencing procedures related to the § 3553(a) factors did not require plain error review because defendant raised the relevance of those factors in its sentencing memorandum and at the sentencing hearing, so that he was “not require[d] ... to re-raise them”). Further, even if we agreed with our colleague that the plain error standard of review applied, we would nevertheless find that the District Court’s failure to calculate a final guidelines range — leaving us unable to review the procedural and substantive bases of the sentence — is an error that is plain, that affects the substantial rights of the parties, and that could “seriously affect the fairness, integrity or public reputation of judicial proceedings.” United States v. Vazquez-Lebron, 582 F.3d 443, 446 (3d Cir.2009) (internal quotation omitted); id. at 446-47 (finding plain error where the District Court “did not accurately follow the second and third steps of the procedure set out in [United States v. Gunter, 462 F.3d 237, 247 (3d Cir.2006) ],” and thus we could not “know the District Court’s intention in sentencing [the defendant]”). A. Loss calculation The parties dispute a number of the calculations that went into the District Court’s determination of the loss attributable to Fumo’s fraud. Ultimately, the District Court’s decisions resulted in a loss calculation for Fumo which fell just short of $2.5 million, the threshold for increasing the offense level. “The appropriate standard of review of a district court’s decision regarding the interpretation of the Sentencing Guidelines, including what constitutes ‘loss,’ is plenary. Factual findings, however, are simply reviewed for clear error.” United States v. Napier, 273 F.3d 276, 278 (3d Cir.2001) (internal citation omitted). 1. The Pennsylvania State Senate a. Overpayment of Senate employees Fumo arranged to have a number of Senate employees under his control classified at higher salary grades than they were entitled to be based on their duties and qualifications. In order to calculate the losses attributable to this fraud, the Government reviewed the human resources manual to determine the proper classification for each employee based on testimony about the work they actually performed and then calculated the loss to the Senate as the difference between the highest salary each could possibly have been entitled to and the salary each actually received, for a total of approximately $1 million. At the sentencing hearing, Fumo did not dispute the type of work the employees actually performed or the salaries that they actually received. Instead, he argued that the calculations were too speculative because the Chief Clerk of the Senate could not confirm them and because the Senate had failed to fire or reclassify these employees after the fact, implying that the original classifications were somehow justified. Agreeing with Fumo, the District Court excluded the Government’s proposed loss altogether. Of course, the Government bears the burden of establishing, by a preponderance of the evidence, the amount of loss. United States v. Jimenez, 513 F.3d 62, 86 (3d Cir.2008). However, although “the burden of persuasion remains with the Government, once the Government makes out a prima facie case of the loss amount, the burden of production shifts to the defendant to provide evidence that the Government’s evidence is incomplete or inaccurate.” Id. In making a loss calculation, “[t]he court need only make a reasonable estimate of the loss.” United States v. Ali, 508 F.3d 136, 145 (3d Cir.2007) (quoting U.S.S.G. § 2B1.1, Application Note 3(C)). Here, the Government made out a prima facie case of the loss amount, and in response Fumo made only the most minimal showing of “inaccuracy” in the Government’s calculations. In fact, Fumo never really challenges the substance of the Government’s calculations, instead relying on surrounding circumstances to cast speculative doubt on them. Yet it is not surprising that the Chief Clerk of the Senate, who had not reviewed in detail the evidence concerning each employee’s duties, declined to take a position on the stand as to the accuracy of the Government’s calculations. And the Senate’s decision not to reclassify certain of the employees involved could have been prompted by any manner of reasoning or purposes. Although it is possible that the Government made errors in the course of its calculations, there is no reason to think that its figure was not a “reasonable estimate” of the loss, established by a preponderance of the evidence. Accordingly, after reviewing the District Court’s grounds for rejecting the Government’s prima facie showing of the loss amount, we are left with “the definite and firm conviction that a mistake has been committed.” United States v. Grier, 475 F.3d 556, 570 (3d Cir.2007) (en banc) (internal quotation omitted). Further, because the difference in the loss would place Fumo into a higher offense level, the error was not harmless. b. Rubin’s “no-work” contract The Government next objects to the District Court’s decision to exclude from the loss calculation a $150,000, five-year contract awarded to Arnao’s husband Rubin, for which he purportedly performed no services. At the July 8 sentencing hearing, Fumo informed the court that he had gathered additional evidence demonstrating that Rubin had, in fact, completed work under the contract. He submitted the evidence on July 13. The additional material consisted largely of credit card bills and calendar entries, documenting that Rubin had met with people, but not what those meetings had been about. The Government argued that the evidence submitted by Fumo was weak or irrelevant, and noted that Fumo’s current theory that Rubin had worked directly with Fumo and met with people on his behalf contradicted Rubin’s testimony at trial, that the contract was with Rubin’s company, B & R Services, for court services. The District Court declined to rule on the issue of loss from Rubin’s contract, stating that “because of the complexity of the Rubin loss argument in light of the defense submissions, I felt I could not properly resolve it before sentencing. Rather than postpone the sentencing, I declined to rule on it.” (Sealed App. 184-85). This was an abuse of discretion. The Federal Rules require a Court to rule on any disputed matters at sentencing unless “a ruling is unnecessary ... because the court will not consider the matter in sentencing.” Fed.R.Crim.P. 32(i)(3)(B). Fumo argues that, because the court excluded the $150,000 from its loss calculation, it did not “consider the matter in sentencing,” and thus