OPINION OF THE COURT

PADOVA, Senior District Judge.

James and Theresa DeMuro ("the DeMuros") owned TAD Associates, LLC ("TAD"), an engineering and surveying company in New Jersey. Between 2002 and 2008, TAD failed to pay over to the IRS more than half a million dollars in taxes that it had withheld from its employees' paychecks. As the owners and managers of TAD, the DeMuros were convicted after a jury trial of conspiracy to defraud the United States, in violation of 18 U.S.C. § 371, and 21 counts of failure to account and pay over employment taxes, in violation of 26 U.S.C. § 7202. The DeMuros argue on appeal that the District Court committed numerous evidentiary errors at trial, and further erred at sentencing by applying a two-level increase to their offense levels for abuse of a position of trust, pursuant to U.S.S.G. § 3B1.3. For the following reasons, we affirm the convictions, but remand for resentencing.

I.

Appellants James DeMuro ("James") and Theresa DeMuro ("Theresa"), husband and wife, were the sole owners and managers of TAD. James is an engineer and was responsible for the business's surveying operations, while Theresa was the office manager and directed TAD's marketing efforts.

Between 2002 and 2008, TAD withheld from its employees' paychecks trust fund taxes, i.e., taxes that the employees owed to the federal government for Social Security and Medicaid, and as income tax. Employers typically withhold these taxes from their employees' paychecks, and then hold them in trust until it is time to pay them over to the United States. However, over the course of 21 calendar quarters between 2002 and 2008, TAD failed to pay over to the IRS a total of $546,247 in trust fund taxes that it had collected from its employees.

IRS Revenue Officer Michelle Hornby first contacted the DeMuros in 2002 about the trust fund taxes that TAD owed. Hornby informed the DeMuros that they could be held personally liable for the unpaid taxes because they were responsible for paying TAD's taxes. Hornby also advised them that the IRS considered the trust fund taxes to be TAD's top priority, and that TAD was not permitted to spend the withheld trust fund taxes for any other purpose. From 2002-2005, the DeMuros expressed interest in paying the back taxes and ostensibly worked with Hornby in an effort to pay off the debt. At the same time, however, the DeMuros were spending large amounts of money on personal expenditures and seemingly discretionary business expenses.

In October 2003, the IRS assessed the DeMuros personally with a trust fund recovery penalty, pursuant to 26 U.S.C. § 6672. The amount of the penalty was the amount that TAD owed in trust fund taxes. Hornby again warned the DeMuros that they needed to start making timely deposits of TAD's trust fund taxes, but they did not. As a result, Hornby required TAD to increase the frequency of its payments of taxes to the IRS from quarterly to monthly.

In June 2004, when the DeMuros were still failing to make timely deposits of the trust fund taxes, the IRS required TAD to establish a special trust account in favor of the United States, pursuant to 26 U.S.C. § 7512. The DeMuros were required to put withheld trust fund taxes into this account within two days of withholding the money, and to keep the funds in the account until they were paid to the IRS. However, after paying funds into the trust account, the DeMuros withdrew some of these funds to pay unrelated expenses. They later closed the account without IRS permission.

On July 15, 2010, a grand jury indicted the DeMuros on 22 counts: one count of conspiracy to defraud the United States, 18 U.S.C. § 371, and 21 counts of failure to account and pay over employment taxes, 26 U.S.C. § 7202, one count for each quarter of unpaid taxes. At trial, the DeMuros conceded that they did not pay over the $546,242 that TAD owed to the IRS in trust fund taxes, but argued that their failure to pay TAD's taxes was not willful. In support of this position, they emphasized that they negotiated with the IRS in an attempt to pay off the delinquent taxes over time, and pointed to evidence that they even attempted to refinance properties to obtain funds to pay the delinquent taxes. Theresa also asserted an innocent spouse defense at trial, arguing that James, not she, was responsible for collecting and paying over TAD's taxes.

The Government responded to these arguments by, inter alia, presenting evidence that the DeMuros spent lavishly on personal and discretionary business items while professing an inability to pay the delinquent taxes. The evidence showed that, in the 21 quarters in which TAD failed to pay over trust fund taxes in the amount of $546,242, the DeMuros spent $5,043,867 from TAD and personal accounts. To prove that Theresa was responsible for paying the taxes, the Government presented evidence that Theresa had signature authority over all of TAD's bank accounts, freely spent TAD's money, and had the power to hire and fire employees.

The jury convicted both James and Theresa on all 22 counts. At sentencing, the District Court applied a two-level enhancement to the DeMuros' offense levels for abuse of a position of trust, explaining that the DeMuros were in a position of trust vis-a-vis the IRS on account of the special trust fund account, and that the DeMuros abused that position of trust by using trust fund money for expenses other than trust fund taxes. The District Court sentenced both James and Theresa to 51 months' imprisonment. These timely appeals followed.1

II.

On appeal, the DeMuros raise numerous challenges to the District Court's decisions admitting or excluding evidence. We review evidentiary rulings for abuse of discretion. United States v. Friedman, 658 F.3d 342, 352 (3d Cir.2011) (citing United States v. Starnes, 583 F.3d 196, 213-14 (3d Cir.2009)). Evidentiary errors objected to at trial are reviewed for harmless error. Fed.R.Crim.P. 52(a); Neder v. United States, 527 U.S. 1, 7, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999). An error that was not objected to at trial is reviewed for plain error. Fed.R.Crim.P. 52(b).

"An error in an evidentiary ruling is harmless error when `it is highly probable that the error did not affect the result.'" Friedman, 658 F.3d at 352 (quoting Hill v. Laeisz, 435 F.3d 404, 420 (3d Cir.2006)). "`Plain error exists only when (1) an error was committed (2) that was plain, and (3) that affected the defendant's substantial rights.'" United States v. Lopez, 650 F.3d 952, 961 (3d Cir.2011) (quoting United States v. Lessner, 498 F.3d 185, 192 (3d Cir.2007)). An error is "plain" when it is "clear or obvious." United States v. Russell, 134 F.3d 171, 180 (3d Cir.1998) (citation omitted). Where a plain error has occurred, we exercise discretion in determining whether that error warrants reversal. United States v. Campbell, 295 F.3d 398, 404 (3d Cir.2002). We will reverse the district court "only if the error `seriously affect[s] the fairness, integrity, or public reputation of judicial proceedings.'" United States v. Stevens, 223 F.3d 239, 242 (3d Cir.2000) (quoting United States v. Olano, 507 U.S. 725, 732, 113 S.Ct. 1770, 123 L.Ed.2d 508 (1993)).

A.

The DeMuros first argue that the District Court erred in admitting much of the Government's evidence that was intended to establish the DeMuros' willfulness and responsibility for paying TAD's taxes, two of the elements the Government must prove to obtain a conviction under 26 U.S.C. § 7202.2 See Cheek v. United States, 498 U.S. 192, 199-200, 111 S.Ct. 604, 112 L.Ed.2d 617 (1991) (citations omitted); United States v. Thayer, 201 F.3d 214, 219 (3d Cir.1999). Willfulness in a tax evasion case is "a voluntary, intentional violation of a known legal duty." Cheek, 498 U.S. at 201, 111 S.Ct. 604 (citations omitted). Where a defendant proves that he had a good faith belief that he was not violating the tax code, regardless of whether that belief was objectively reasonable, he has established that he did not act willfully. Id. at 201-03, 111 S.Ct. 604.

As mentioned above, the Government presented a variety of evidence to prove willfulness and responsibility, including evidence of the DeMuros' personal spending, Theresa's handling of her father's assets, and Theresa's role in the resignation of TAD employee Joel Boriek. The DeMuros argue that the District Court abused its discretion in admitting this evidence because the evidence was not only irrelevant to any issue at trial, but also overly prejudicial.

1.

Several witnesses testified at trial about the DeMuros' personal spending habits, including their luxury vacations, nice homes, and Theresa's substantial home shopping network expenditures. The Government also presented evidence that the DeMuros wrote checks on TAD's accounts to pay for some of their personal expenditures. The DeMuros contend that the District Court abused its discretion not only by finding this evidence to be relevant, but moreover by not finding that the voluminous evidence of their personal spending was overly prejudicial compared to any minimal relevance it may have had to willfulness.

a. Relevance

Contrary to the DeMuros' assertion, evidence of personal spending can be relevant evidence in criminal tax actions and was, in fact, relevant here. We find guidance in two cases from our sister circuits. In United States v. Ellis, 548 F.3d 539, 543 (7th Cir.2008), the United States Court of Appeals for the Seventh Circuit considered a defendant whose "principal defense" to a tax evasion charge was that she "was too busy to notice or remember her tax obligations." Under those circumstances, "the ways [defendant] was spending her time— traveling to Florida, buying cars, purchasing and overseeing the decoration of her two million dollar home—were relevant" because they rebutted her assertion that she was "too busy" to pay her taxes. Id. The court further noted that the same evidence of personal spending was also relevant to undermine a defense that the defendant "simply forgot her tax liability, since most people would inquire as to why they have an unexpected additional two million dollars to spend on themselves." Id.

The United States Court of Appeals for the Sixth Circuit has also addressed the use of evidence of personal spending in a criminal tax case. United States v. Blanchard, 618 F.3d 562 (6th Cir.2010). The court stated that:

Id. at 569-70. In Blanchard, the defendant argued that he lacked the ability to pay the taxes. The court therefore held that the fact that the defendant had spent money on gambling and other personal pleasures was relevant to negating his defense. Id. at 570 n.4.

We adopt the rationale of Ellis and Blanchard insofar as they stand for the proposition that evidence of personal spending can be relevant to rebut a defendant's defense that he has not acted willfully. Of course, whether such evidence is relevant in any particular case will be dependant on a case-specific inquiry.

Employing these principles in the case at bar, we conclude that the District Court did not abuse its discretion in finding the evidence of the DeMuros' personal spending to be relevant to the jury's assessment of willfulness in light of the DeMuros' defensive arguments at trial. The DeMuros argued at trial that they were genuinely trying to pay off their taxes and, thus, were not acting willfully in not paying them. Given this argument, evidence that the DeMuros spent money on vacations and other personal items—money that they could have used to pay their taxes—belies their assertions that they were sincerely attempting to pay their back taxes as expeditiously as possible. Thus, the District Court did not abuse its discretion in concluding that the evidence of personal spending was relevant proof that the DeMuros' failure to pay TAD's taxes was willful.

The evidence of the DeMuros' lavish personal spending was also relevant to the conspiracy charge as it was probative of their conspiratorial purpose. See United States v. Boria, 592 F.3d 476, 481 (3d Cir.2010) ("To establish a charge of conspiracy, the Government must show . . . a shared unity of purpose. . . .") (citations omitted). The Government's theory at trial was that the DeMuros failed to pay their taxes because they wanted to live a comfortable lifestyle and, thus, they agreed to evade the tax laws so that they would have money to spend on themselves. Given this theory, the evidence that they did live a relatively lavish lifestyle, and how much money they spent on that lifestyle, was relevant.

Finally, some of the evidence of the DeMuros' personal spending was also relevant to whether they were the parties responsible for paying TAD's taxes. Evidence of an employee's "`ability to sign checks on the company's bank accounts'" and evidence of a company's "`payment of other creditors in lieu of the United States'" are relevant to determining whether a particular individual is responsible for paying over withholding taxes. Greenberg v. United States, 46 F.3d 239, 243 (3d Cir.1994) (quoting Brounstein v. United States, 979 F.2d 952, 954-55 (3d Cir.1992)). In the case at bar, the evidence showed that the DeMuros wrote checks on TAD accounts to pay for some of their personal expenses. This evidence was relevant, regardless of the fact that it involved personal spending, because it showed both that James and Theresa had the ability to sign TAD checks and that they chose to pay other creditors in lieu of the United States. Thus, the evidence of the DeMuros' personal spending was relevant to their willfulness and responsibility, as well as the conspiracy charge.

b. Prejudice

Having addressed the issue of relevance, we turn to the DeMuros' argument that the evidence was overly prejudicial and that the District Court therefore abused its discretion in admitting it. According to the DeMuros, the extensive evidence of their personal expenditures was prejudicial because it fomented class prejudice by depicting a life of luxury that came at the expense of the DeMuros' employees and the American people.

The District Court may exclude relevant evidence "if its probative value is substantially outweighed by a danger of. . . unfair prejudice, confusing the issues, misleading the jury, undue delay, wasting time, or needlessly presenting cumulative evidence." Fed.R.Evid. 403. "The District Court's discretion is `construed especially broadly in the context of Rule 403.'" Friedman, 658 F.3d at 355 (quoting United States v. Kemp, 500 F.3d 257, 295 (3d Cir.2007)). Furthermore, "`Rule 403 does not provide a shield for defendants who engage in outrageous acts, permitting only the crimes of Caspar Milquetoasts to be described fully to a jury. It does not generally require the government to sanitize its case, to deflate its witnesses' testimony, or to tell its story in a monotone.'" United States v. Cross, 308 F.3d 308, 325 (3d Cir.2002) (quoting United States v. Gartmon, 146 F.3d 1015, 1021 (D.C.Cir. 1998)).

As noted above, the evidence of the DeMuros' personal spending was relevant to their responsibility for paying TAD's taxes, their willfulness, and the conspiracy charge. Thus, its probative value was significant. Further, despite the DeMuros' contention that this evidence was overwhelming and voluminous at trial, the evidence of personal spending was largely limited to one witness, IRS Agent Carlos Natasi, with other witnesses only briefly mentioning the DeMuros' spending habits. While we are sensitive to the effect that evidence of a defendant's liberal spending habits can have on a jury, particularly in these lean economic times, the evidence admitted in this case, i.e., evidence of vacations, jewelry, cars, and parties, was not so inflammatory as to carry a great risk of prejudice. It certainly was not so prejudicial as to substantially outweigh its very significant probative value. Cf. Blanchard, 618 F.3d at 570 (allowing evidence of gambling expenses to prove willfulness, despite the risk that some jurors may have negative views towards gambling). Accordingly, we conclude that the District Court did not abuse its discretion in allowing Agent Natasi and others to testify about how the DeMuros spent money.

2.

The DeMuros next argue that the District Court erred in permitting IRS Agent Carlos Natasi to testify regarding Theresa's handling of her ailing father's assets. According to Agent Natasi, Theresa received power of attorney over her father because he was mentally incompetent. Agent Natasi testified that Theresa used that power of attorney to mortgage her father's home for approximately $300,000, then placed the money into her personal bank account and spent some of the proceeds on personal items, such as jewelry. The DeMuros argue that this evidence was not only irrelevant, but also overly prejudicial because it suggested that Theresa took advantage of her ailing father.

Although this is a closer call than some of the personal spending evidence, we find that the District Court did not abuse its discretion in admitting this evidence. As explained above, how the DeMuros spent their money was relevant to willfulness in light of their defensive arguments at trial. More particularly, the fact that Theresa came upon nearly $300,000 and spent that money on personal items undercuts her defense that she and James were attempting to work out a payment plan with the IRS. Put another way, Theresa's choice to spend this money on personal items rather than the delinquent taxes reflects that she regarded her known obligation to pay the trust fund penalty with little urgency. We therefore find this evidence to be probative of willfulness.

While we recognize that the source of the $300,000 windfall, i.e., the mortgage on Theresa's ailing father's house, is not itself relevant to willfulness, the Government was not required to disentangle this aspect of the story when making its case to the jury. See Fed.R.Evid. 401 advisory committee's note (explaining that background information is relevant). The DeMuros maintain that the evidence was unduly prejudicial because the jury undoubtedly inferred that Theresa effectively stole this money from her mentally incapacitated father. However, the DeMuros rebutted any such inference with evidence that Theresa was her father's sole heir, and thus was destined to inherit the house upon his death. Indeed, Theresa's counsel argued this point to the jury in closing arguments, saying that the $300,000 was essentially Theresa's money anyway. Accordingly, any potential prejudicial effect of the evidence was mitigated and any residual prejudice was surely outweighed by the evidence's probative value. We therefore find that the District Court did not abuse its discretion in admitting this evidence.

3.

The DeMuros also challenge the District Court's decision to admit evidence about the circumstances surrounding TAD employee Joel Boriek's resignation. Boriek testified that he was working four days a week at TAD because his wife was ill, but then Theresa told him he needed to work five days a week, and he quit rather than change his schedule. The evidence at trial was that Theresa told another employee that she was concerned about the medical benefits TAD was paying to Boriek, so she asked him to work five days a week knowing that he would rather quit. The DeMuros argue that testimony about Theresa's motivations in this regard was irrelevant and overly prejudicial. They therefore contend that the evidence should have been excluded.

Contrary to the DeMuros' arguments, the evidence regarding Theresa's involvement in Boriek's situation was relevant to the issues at trial. First, the evidence was probative of the fact that Theresa had the power to discharge Joel Boriek, which is directly relevant to whether she was responsible for paying TAD's taxes. See Greenberg, 46 F.3d at 243 (citation omitted) (explaining that an individual is responsible to pay his employer's taxes if he has significant control over the employer's finances, and stating that one factor a court may consider in assessing that control is whether the individual is in charge of hiring and firing employees). Second, evidence that Theresa wanted Boriek to quit so as to save money on medical benefits was probative of her control over TAD's financial affairs, which is also relevant to establishing her responsibility for paying TAD's taxes. Id. Significantly, Theresa vehemently argued that she was not responsible for paying TAD's taxes. Given that argument, the evidence of her authority to discharge employees and her control over TAD's finances was highly probative of a contested issue. While the DeMuros correctly assert that this evidence was not flattering to Theresa, we do not find it so prejudicial as to substantially outweigh its probative value. Accordingly, we conclude that the District Court did not abuse its discretion in allowing evidence regarding the circumstances of Joel Boriek's separation from TAD.

B.

The DeMuros challenge the admission of a portion of the testimony of enrolled agent Victor Minelli. An enrolled agent is one of three professionals who can represent a taxpayer before the IRS, along with CPAs and attorneys. In that representative capacity, an enrolled agent is required to have expertise in tax issues. Minelli testified that the DeMuros hired him in 2005 to represent them before the IRS with respect to TAD's tax problems. He testified that he reviewed two ongoing appeals that the DeMuros had filed challenging decisions of the IRS, and that he determined that the appeals had no merit and should be withdrawn. The DeMuros argue that his testimony that these two tax appeals were meritless constituted expert opinions, which the District Court should not have permitted without first finding Minelli to be qualified as a expert. The DeMuros did not object to this testimony at trial, so we apply plain error review.

The Federal Rules of Evidence limit the opinions that a lay witness may provide. Rule 701 provides that lay opinions must be "(a) rationally based on the perception of the witness, (b) helpful to a clear understanding of the witness' testimony or the determination of a fact in issue, and (c) not based on scientific, technical, or other specialized knowledge within the scope of Rule 702." Fed.R.Evid. 701. It is not uncommon for witnesses who testify about tax consequences to be qualified as experts. See United States v. Stadtmauer, 620 F.3d 238, 269 (3d Cir. 2010) (listing cases where IRS agents were qualified as experts and provided opinions as to the tax consequences of a transaction). On the other hand, we have allowed professionals to give lay opinions when the opinions are based on personal knowledge of the issues, along with specialized experience. See Eisenberg v. Gagnon, 766 F.2d 770, 781 (3d Cir.1985) (allowing lay opinion of lawyer, who specialized in business litigation and securities law, about what information should have been included in a private offering memorandum based on his personal knowledge of the documents).

In the case at bar, the first appeal about which Minelli testified involved a lien the IRS had filed against the DeMuros' properties, which the DeMuros appealed. Minelli testified that, under the law, if a person owes taxes, the IRS can file a lien. He further testified that there was no dispute as to whether the DeMuros owed the taxes for which the IRS lien was sought and, thus, the DeMuros' appeal had no merit. Even if we assume arguendo that this opinion was an expert opinion, we conclude that it was not "plain error" for the District Court to admit it because it is not "clear or obvious" that this opinion was based on any specialized knowledge of the tax code, rather than Minelli's personal knowledge and specialized expertise. Russell, 134 F.3d at 180 (citation omitted). Moreover, we cannot conclude that such error "`seriously affect[ed] the fairness, integrity, or public reputation of judicial proceedings.'" Stevens, 223 F.3d at 242 (quoting Olano, 507 U.S. at 732, 113 S.Ct. 1770).

The second appeal about which Minelli testified concerned the DeMuros' appeal of the amount of their trust fund penalty. The DeMuros argued in the appeal that the IRS's calculation did not reflect some payments they had made. Minelli testified at trial that he checked the records and that the IRS had credited all of the DeMuros' payments. He therefore opined that the DeMuros' appeal was meritless. We conclude that any error in allowing this testimony was not "plain" because it is not "clear or obvious" that this opinion was based on specialized knowledge of the tax code, as opposed to Minelli's experience in reading IRS account statements and the application of simple mathematical concepts. Accordingly, we find that the District Court did not commit reversible error when allowing Minelli to opine that the DeMuros' two appeals had no merit.

C.

The DeMuros argue that the District Court erred in permitting the Government to present evidence that TAD withheld money from employee paychecks for health insurance, retirement, and child support payments, and then failed to pay over these funds to their intended recipients.3 The District Court admitted this evidence pursuant to Rule 404(b), as evidence of the DeMuros' intent, plan, and motive.

Rule 404(b) provides that evidence of other bad acts, although not admissible to show character, are admissible to prove "motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident." Fed. R.Evid. 404(b). The Supreme Court has established a four-part test to determine whether evidence was properly admitted under Rule 404(b): 1) the evidence must have a proper purpose under 404(b); 2) the evidence must be relevant; 3) the probative value of the evidence must outweigh its potential for prejudice; and 4) the court must instruct the jury to consider the evidence only for the limited purpose for which it is admitted. United States v. Sampson, 980 F.2d 883, 886 (3d Cir.1992) (citing Huddleston v. United States, 485 U.S. 681, 691-92, 108 S.Ct. 1496, 99 L.Ed.2d 771 (1988)). When offering such evidence, the Government "`must clearly articulate how th[e] evidence fits into a chain of logical inferences, no link of which can be that because the defendant committed... [such an act] before, he therefore is most likely to have committed this one.'" United States v. Lee, 612 F.3d 170, 186 (3d Cir.2010) (second alteration in original) (quoting Sampson, 980 F.2d at 887). Rule 404(b) "is inclusive, not exclusive, and emphasizes admissibility." Sampson, 980 F.2d at 886 (citation omitted).

Under the Supreme Court's four-part test, we consider in steps one and two whether the evidence was relevant to a proper purpose under Rule 404(b). As noted above, the District Court below admitted the evidence that TAD did not pay over withheld child support payments, health insurance premiums, and retirement monies based on the Government's theory that the evidence was relevant to intent, motive, and plan. We agree that the evidence of failure to pay over other withheld monies was relevant to the DeMuros' intent, as it tended to prove that the DeMuros' failure to pay over the trust fund taxes was not unintentional or the result of a misunderstanding or mistake. Accord United States v. Daraio, 445 F.3d 253, 264 (3d Cir.2006) (allowing evidence of prior failures to pay taxes to prove willfulness in a tax evasion case). The evidence was also probative of a pattern, plan, and motive of the DeMuros to surreptitiously take money from TAD employees by way of legitimate withholdings and then retain the money for their own benefit. We therefore conclude that the evidence was relevant to several proper purposes under Rule 404(b).

In the next step in the Supreme Court's test, we weigh the probative value of the evidence against its potential for prejudice. Here, the above-noted probative value of the challenged evidence was significant. Moreover, we agree with the District Court that the evidence was not unduly prejudicial. Accordingly, we conclude that the probative value of this evidence outweighed its potential for prejudice. Cf. Daraio, 445 F.3d at 264 (finding that the probative value of the evidence of past failures to pay taxes outweighed the potential for prejudice even though the past conduct was the same as the charged conduct).

Finally, in step four of the four-part test, we consider whether the lower court gave a proper jury instruction informing the jury of the limited purposes of the evidence. Here, the District Court properly instructed the jury that the evidence could be considered as proof of state of mind, motive, opportunity, pattern, and lack of mistake, all of which are proper uses of the evidence under the circumstances presented.4 Consistent with this instruction, the Government argued during its closing that the evidence showed motive and state of mind. Accordingly, the Government and the District Court both explained to the jury the proper use of this evidence. We therefore conclude that the District Court did not abuse its discretion in admitting the evidence of the DeMuros' failure to pay over other withheld money.

D.

Theresa objects to the District Court's failure to strike the Government's closing argument reference to the DeMuros' prior bad acts of withholding money from their employees and not paying it over. Theresa contends that the Government impermissibly argued that this evidence showed the DeMuros' bad character and propensity to steal withheld money. The DeMuros did not object to this part of the closing argument at trial, so we review it for plain error.

In its rebuttal argument, the Government argued to the jury:

(A363 (emphasis added).)

As discussed above, the evidence that the DeMuros effectively stole money from their employees by using money withheld for other purposes for their own personal gain was admitted to show intent, among other proper usages, pursuant to Rule 404(b). The Government here properly argued that this evidence was related to the DeMuros' intent, i.e., their lack of good faith. Furthermore, the juxtaposition of the Government's statement with its description of the DeMuros' expenditures links the 404(b) evidence to motive, another proper purpose under Rule 404(b). Accordingly, it was not plain error for the District Court to allow the Government to make this argument to the jury.

E.

The DeMuros argue that the District Court improperly excluded four pieces of testimony from IRS Agent Hornby regarding their negotiations with the IRS. In their view, the excluded evidence would have shown that they were trying to pay their taxes and, thus, not acting willfully. They also argue that this evidence showed that they had a good faith belief that their conduct was not criminal.

The first piece of excluded testimony at issue is certain testimony from Agent Hornby that the DeMuros attempted to finance their properties but were unable to due to IRS liens. The DeMuros argue that this evidence was probative of the fact that they did not willfully fail to pay their taxes because it shows that they were attempting to raise money to pay the taxes through refinancing. However, contrary to the DeMuros' suggestion, the exclusion of this particular testimony did not deprive the jury of important probative information because Hornby testified at length about the DeMuros' efforts at refinancing, and the DeMuros actually argued to the jury in closing arguments that their refinancing efforts proved that they did not act willfully. Thus, there was ample evidence of the DeMuros' attempts to refinance and we reject the DeMuros' argument that the District Court abused its discretion in excluding this additional testimony.

The DeMuros also challenge the exclusion of a note that Hornby wrote after a meeting with James, in which Hornby stated that she had advised James that the IRS could shut down TAD. The District Court excluded this evidence on the grounds of confusion, delay, and minimal relevance. The DeMuros argue that this evidence should have been admitted as probative of their lack of willfulness, because no reasonable person would willfully fail to pay their taxes knowing that the IRS could shut down their business. However, the note at issue would have been duplicative of other testimony, as Hornby had already testified that she had told James about the possibility of shutting down the business. Moreover, even in the absence of the testimony about the note, the DeMuros argued to the jury in closing that they knew of the potential shutdown of the business and would not have willfully failed to pay their taxes risking such a shutdown. Under these circumstances, it was not an abuse of discretion for the District Court to exclude Hornby's merely redundant note.

Third, the DeMuros argue that the District Court erred by excluding any testimony that the reason that the negotiations between the IRS and the DeMuros ceased was that the IRS referred the case for criminal prosecution. In the DeMuros' view, the exclusion of this evidence left the jury free to speculate that the DeMuros themselves chose to stop negotiating with the IRS, which undermined their lack of willfulness defense. The District Court, however, excluded this evidence because it opened the door to jury nullification, by inviting the jury to reason that the IRS should have continued to pursue the matter civilly rather than criminally. See United States v. Buras, 633 F.2d 1356, 1360 (9th Cir.1980) (explaining that the availability of civil remedies in a tax case "is irrelevant to the issue of criminal liability" and finding that a jury instruction regarding civil remedies "would serve only to confuse the jury"). Significantly, the DeMuros concede that this evidence would have opened the door to jury nullification. Moreover, we believe it unlikely that the jury would speculate that the DeMuros terminated the negotiations with the IRS without any evidence regarding the circumstances of the termination and based solely on the fact that the negotiations ended. We thus conclude that it was not an abuse of discretion for the District Court to find that the concern about jury nullification outweighed the risk that the lack of this evidence would give rise to a speculative negative inference.

Fourth, the DeMuros challenge the District Court's decision to exclude Hornby's testimony that she never told them that their case had been referred for criminal prosecution or that their conduct was criminal. The DeMuros argue that this evidence was probative of whether they knew that their conduct was criminal and, thus, should have been admitted. This evidence, however, had only minimal probative value. The mere fact that Hornby did not tell the DeMuros that their conduct was criminal is not strong evidence that the DeMuros honestly believed that their conduct was legal. Further, as the District Court stated, this evidence had the potential for confusion and opened the door to jury nullification. We thus conclude that the District Court did not abuse its discretion by excluding it.

Finally, the DeMuros argue that these four rulings in combination effectively took their good faith defense away from the jury. While it is improper to exclude evidence of a good faith defense based solely on a judicial determination that the good faith belief lacks credibility, it is "proper to exclude evidence having no relevance or probative value with respect to willfulness." Cheek, 498 U.S. at 203, 111 S.Ct. 604. Here, there is no basis in the record to conclude that the District Court excluded any evidence relating to good faith due to a determination that the defense lacked credibility. Rather, the evidence was excluded because it was irrelevant, cumulative, and/or because it was outweighed by a concern for confusion. Accordingly, we find that the District Court did not improperly take the DeMuros' good faith defense away from the jury.

F.

Theresa challenges the District Court's decision to exclude the testimony of IRS Agent Lloyd Walling. Walling was an IRS agent assigned to investigate the DeMuros' former company, DA Resources, which had also failed to pay over trust fund taxes. Theresa called Walling to testify about the contents of a written statement made by James, in which James stated that only he, and not Theresa, was responsible for paying DA Resources' taxes. The District Court excluded this testimony on the grounds that 1) it was irrelevant to whether Theresa was responsible for paying TAD's taxes, 2) it had the potential to confuse the jury, and 3) James' statement was inadmissible hearsay. As a result of this ruling, Walling did not testify at all. Theresa argues that Walling's testimony was relevant and admissible either as a non-hearsay party admission or under the hearsay exception for statements against interest.

Even assuming arguendo that Theresa is correct that this evidence was admissible, we conclude that exclusion of the evidence was harmless. First, the evidence against Theresa with respect to her control and responsibility was substantial: she was an owner and manager of TAD, she had signatory authority over all of TAD's accounts, she signed numerous checks drawn on TAD's accounts during the relevant period, and she had the power to hire and fire employees. See Greenberg, 46 F.3d at 243 (citation omitted) (listing the factors for determining whether an individual was responsible for paying a company's taxes). Second, the testimony of Walling was of minimal probative value because it related to a former company of the DeMuros, not TAD. Moreover, James made his statement at a time when his responsibility to pay DA Resources' taxes was clear, but Theresa's was not, meaning that the statement was self-serving in spite of appearing to be a statement against interest. Accordingly, we conclude that "`it is highly probable that the error did not affect the result'" of the trial, and thus the error was harmless. Friedman, 658 F.3d at 352 (quoting Hill, 435 F.3d at 420).

Theresa also argues that the District Court's decision to exclude Walling's hearsay testimony effectively excluded any other testimony that Walling may have given regarding Theresa's responsibilities with respect to DA Resources. She argues that, as a result of the exclusionary ruling, she was prevented from presenting a defense in violation of her Fifth and Sixth Amendment rights. Theresa, however, did not proffer any further testimony from Walling at trial, and it is not clear from the record what the substance of his testimony would have been. Thus, Theresa cannot now raise this issue on appeal. See Fed. R.Evid. 103(a)(2). Moreover, even on appeal, Theresa has not once described the substance of any further testimony from Walling. Therefore, we have no basis upon which to conclude that the District Court erred by foreclosing any further testimony of Agent Walling.

Accordingly, we affirm the convictions of both James and Theresa DeMuro.5

III.

At sentencing, the District Court applied a two-level enhancement to the DeMuros' base offense levels for abuse of position of trust, pursuant to U.S.S.G. § 3B1.3. The District Court applied the enhancement based on the DeMuros' conduct with respect to the trust fund account imposed by the IRS, into which the DeMuros were obliged to pay TAD's trust fund taxes within two days of withholding the taxes. The DeMuros, however, made several withdrawals from this account for other business purposes, and closed the account without IRS permission. We review whether a defendant occupied a position of trust de novo. United States v. Hart, 273 F.3d 363, 376 (3d Cir.2001) (citation omitted). We review whether a defendant abused that position of trust for clear error. Id. (citation omitted).

Pursuant to U.S.S.G. § 3B1.3, a two-level increase to a defendant's offense level applies "[i]f the defendant abused a position of public or private trust ... in a manner that significantly facilitated the commission or concealment of the offense." In applying this enhancement, we employ "a two-step analysis: (1) whether the defendant occupied a position of public or private trust; and (2) whether the defendant abused this position of trust in a way that significantly facilitated the crime." Hart, 273 F.3d at 375 (citing United States v. Iannone, 184 F.3d 214, 222 (3d Cir. 1999)).

In United States v. Pardo, 25 F.3d 1187, 1192 (3d Cir.1994), we set forth three factors to consider in addressing the first step in the two-step analysis, stating:

Id. See also Hart, 273 F.3d at 375. The Sentencing Guidelines further explain that a position of trust is "characterized by professional or managerial discretion (i.e., substantial discretionary judgment that is ordinarily given considerable deference)." U.S.S.G. § 3B1.3 cmt. n.1.

"`[T]hese factors should be considered in light of the guiding rationale of [§ 3B1.3]: `to punish `insiders' who abuse their positions rather than those who take advantage of an available opportunity.'" United States v. Dullum, 560 F.3d 133, 140 (3d Cir.2009) (quoting Pardo, 25 F.3d at 1192). Although the enhancement often arises in an employment relationship, we have applied it to other forms of trust relationships. E.g., id. at 140-41 (finding that enhancement applies to a fraud committed by a trusted church adviser); see also Pardo, 25 F.3d at 1190-91 (citing cases involving a mother/daughter relationship and a babysitter/child relationship, among others).

Our inquiry is whether the DeMuros were in positions of trust vis-a-vis the IRS based on their positions as signatories of the trust fund account set up to benefit the IRS.6 In making this determination, we compare the relative position of the DeMuros, as signatories to the trust fund account, to that of other business managers who withhold trust fund taxes. See United States v. Lieberman, 971 F.2d 989, 994 (3d Cir. 1992) (citing United States v. Hill, 915 F.2d 502, 507-08 (9th Cir.1990)) (noting that the inquiry involves comparing the defendant's position relative to other persons who could commit bank embezzlement). In that light, each of the three Pardo factors supports the conclusion that the DeMuros were not in positions of trust based on the trust fund account.

First, we consider whether the trust fund account put the DeMuros in a position to commit a difficult-to-detect wrong. The IRS imposed the trust fund account on the DeMuros so that it would be easier for the IRS to monitor whether the DeMuros were properly paying the trust fund taxes. Indeed, before the creation of the trust fund account, the IRS could only monitor the DeMuros' compliance on a monthly or quarterly basis, whereas afterwards it could monitor the trust fund daily, tracking all deposits into and withdrawals from the account. Under these circumstances, had the IRS been more watchful, it could have discovered rather quickly that the DeMuros were withdrawing money from the account for improper purposes. Thus, the trust fund in no way put the DeMuros in a better position to commit a difficult-to-detect wrong.

Second, we look to the authority given to the DeMuros over the trust fund taxes as a result of the trust fund account. Before the trust fund account, the DeMuros were required to pay the withholding taxes on a quarterly, and later monthly, basis, and could co-mingle the withheld money with other TAD assets. After the establishment of the trust fund, the DeMuros were required to pay the withholdings into the account within two days. Thus, the trust fund account decreased any discretion and authority the DeMuros had with respect to how to hold the money until it was time to pay it to the IRS. Thus, this factor weighs against the imposition of the enhancement.

Third, we consider the extent to which the IRS relied on the integrity of the DeMuros. The IRS imposed the trust fund account on the DeMuros because it could not rely on their integrity. The trust fund account was only set up because the DeMuros had repeatedly failed to pay TAD's taxes in full. The IRS did not trust them to hold the withheld money without spending it. Accordingly, this third Pardo factor also weighs against the imposition of the enhancement. We therefore conclude that the DeMuros were not in a position of trust with respect to the IRS, and that it was error for the District Court to apply the abuse of a position of trust enhancement.

The Government argues, however, that even if the District Court applied the enhancement in error, we should not remand for re-sentencing because the error was harmless. At sentencing, in addition to the abuse of a position of trust enhancement, the District Court applied a two-level increase to the DeMuros' offense level, pursuant to U.S.S.G. §§ 2T1.1 and 2T4.1, because the loss involved was over one million dollars. As a result, each DeMuro had a total offense level of 24 and an advisory Guidelines range of 51 to 63 months' imprisonment. The District Court then varied downwards two levels, because the loss involved just barely triggered the two-level increase, resulting in an adjusted offense level of 22 and a Guidelines range of 41 to 51 months. The District Court sentenced each DeMuro to 51 months' imprisonment, the top of the range.

"[T]he use of an erroneous Guidelines range will typically require reversal." United States v. Langford, 516 F.3d 205, 215 (3d Cir.2008). Under limited circumstances, however, an error in applying the Guidelines can be harmless. Id. "For the error to be harmless, it must be clear that the error did not affect the district court's selection of the sentence imposed." Id. (citing Williams v. United States, 503 U.S. 193, 203, 112 S.Ct. 1112, 117 L.Ed.2d 341 (1992)). The mere fact that the sentence actually imposed falls within both the incorrect Guideline range employed at sentencing and the correct Guideline range is insufficient to show harmless error. Id. at 216, 112 S.Ct. 1112. Rather, "the record must be unambiguous that the miscalculation of the range had no effect" and that "the sentencing judge would have imposed the same sentence under a correct Guideline range." Id. at 216, 217, 112 S.Ct. 1112. "The proponent of the sentence bears the burden of `persuad[ing] the court of appeals that the district court would have imposed the same sentence absent the erroneous factor.'" Id. at 215, 112 S.Ct. 1112 (quoting Williams, 503 U.S. at 203, 112 S.Ct. 1112).

Without the two-level enhancement for abuse of a position of trust, the DeMuros' offense levels would be 22 and their Guidelines ranges would be 41 to 51 months. The Government argues that because this range is the same as the range ultimately used by the District Court, and because the sentences imposed were within this range, the sentencing error was harmless. As discussed, however, the fact that the sentenced imposed falls within the correct range does not alone establish harmless error. Moreover, the Government's argument ignores the fact that the District Court found the Guidelines range to be inappropriate and varied downwards. There is absolutely no basis in the record for us to conclude that the District Court would not have made a similar variance if the abuse of a position of trust enhancement had not applied. We thus cannot conclude that the error here was harmless. We therefore vacate the DeMuros' sentences and remand for re-sentencing.

IV.

For the foregoing reasons, the convictions of the DeMuros are affirmed, but we vacate their sentences and remand for re-sentencing.