Defendants Lewis Tam, Robert Ng, and Co Giang, were part of a car purchasing and insurance fraud scheme involving many participants. All three co-defendants were charged with conspiracy to commit mail fraud and to transport stolen cars in foreign commerce; mail fraud; and transporting stolen cars in foreign commerce. Ng and Tam were also charged with conspiracy to launder money. Following a joint trial, a jury found all three guilty on all counts charged. All three appeal on numerous issues concerning trial and sentencing errors. We have jurisdiction pursuant to 28 U.S.C. § 1291 . We affirm as to all convictions and the sentences of all defendants.

FACTS AND PROCEDURAL HISTORY

Defendants Tam, Ng, and Giang were part of a car purchasing and insurance fraud scheme. The scheme, which took place during 1994 and 1995, was devised by Ken Mak, Norman Kuang, and Kevin Tsuming Yin. Defendants Tam, Ng, and Giang were middle-tier members of the conspiracy who recruited “straw buyers.” The straw buyers would purchase or lease luxury vehicles with a small down payment provided' for them by organizing members of the conspiracy. Car theft and loss insurance were also purchased on each vehicle. After taking possession of the vehicles from the dealership, the straw buyer would turn the car over to the person who had recruited him to make the purchase. The recruiters would then take the car to Los Angeles and deliver it to coeonspirator Yin. Yin would arrange for the shipment of the vehicle, via commercial shipping line, from the port of Los Angeles to Hong Kong. After the car was in transport to Hong Kong, the straw buyer would falsely report the car stolen and file a false theft claim with the insurance company.

The end result of the scheme was that the insurance company would pay a claim to the company that financed the purchase of the vehicle and would reimburse some or all of the down payment money to the straw buyer. This reimbursement was the money the straw buyer was permitted to keep, and thus his or her incentive for making the straw purchase. Once the vehicle reached Hong Kong, the car would then be smuggled into mainland China where it would be sold for two or three times its original value.

To facilitate the scheme, it was necessary to bring money into the United States to pay for the associated costs. This included the money paid to the middlemen/recruiters and straw buyers, the cost of shipping the cars to Hong Kong, as well as money paid to various members of the conspiracy for their roles in the operation. This money was brought into the United States by Ken Mak, primarily by wire transfers from Hong Kong banks. The money was transferred into various banks in Los Angeles and two accounts in Sacramento. The money in the Los Angeles bank accounts was controlled by Ken Mak and Kevin Tsuming Yin. The Sacramento accounts were controlled by defendant Tam and Ng (one account each).

As the scheme evolved, some of the straw buyers recruited others to participate in the scheme both as straw buyers and as middlemen/recruiters. Defendants Tam, Ng, and Giang started out as straw buyers before becoming recruiters. According to the government, Tam and Ng became even more actively involved in the scheme, assuming more prominent positions in the conspiracy. Both Tam and Ng helped transport cars to Los Angeles and acted as conduits for money coming from Ken Mak and Norman Kuang and going to the recruiters and straw buyers.

All told, approximately 112 automobiles were obtained by the conspirators and illegally exported to Hong Kong. Approximately $2,300,000 was wire-transferred from Ken Mak into the various U.S. bank accounts. Additionally, on at least one occasion, Mak physically brought approximately $100,000 of U.S. currency into Los Angeles.

The indictment in this case was filed on July 9, 1998 and charged several people, including defendants Ng, Tam, and Giang with (1) conspiracy to commit mail fraud and to transport stolen cars in foreign commerce ( 18 U.S.C. §§ 371 , 1341, 2314), (2) mail fraud ( 18 U.S.C. § 1341 ), and (3) transporting stolen cars in foreign commerce ( 18 U.S.C. § 2313 ). It also charged defendants Ng and Tam with conspiring to launder money. 18 U.S.C. § 1956 (h).

Defendants Ng, Tam, and Giang went to trial on February 22, 1999. At the close of the government’s case, Ng and Tam made Rule 29 motions to dismiss based on the failure of the indictment to allege an overt act in count ten’s money laundering conspiracy charge. The motion was denied. Following the joint trial, the jury rendered guilty verdicts against all defendants on all counts.

At sentencing, Tam received a term of 97 months; Ng a term of 60 months; and Giang a term of 41 months. Although Ng and Tam’s offense level (29) and criminal history (I) were identical, the district court departed downward four levels for defendant Ng on the basis that he was the sole remaining parent of two small children.

DISCUSSION

I. As to all three defendants:

All three defendants argue that the prosecutor committed structural error in “tipping” the jury as to how to select a foreperson, for example, by recommending a secret ballot. Although the prosecutor’s comments appear inappropriate to us — as they did to the district court, which instructed the prosecutor to refrain from making further comment in this regard— the statements were not, as the defendants urge on appeal, structural error.

The Supreme Court has said that structural errors are those that affect “the framework within which the trial proceeds.” Arizona v. Fulminante, 499 U.S. 279, 310 , 111 S.Ct. 1246 , 113 L.Ed.2d 302 (1991). In contrast, trial errors are those that occur “during the presentation of the case to the jury, and which may therefore be quantitatively assessed in the context of other evidence presented.” Id. at 307-08 , 111 S.Ct. 1246 .

If error was committed in the instant case, it did not rise to the level of a structural defect. In the worst-case see- nario, the jury was improperly influenced as to how it should go about picking a foreperson. But the method suggested— secret ballot — is certainly a permissible one. The defendants’ suggestion that this was an intrusion into the jury’s deliberative function is without case precedent or merit. Furthermore, as trial error, the prosecutor’s comments were in the instant case harmless. The district court appropriately responded to defense counsel’s objection and properly instructed the jury as to the selection and role of the foreperson.

II. As to Tam and Ng:

Tam and Ng argue that the prosecution improperly shifted the burden of proof onto the defendants by thrice stating that the defendant had the same subpoena powers as the government, thereby implying that the defendants had a duty to call witnesses in their own defense.

We review claims of prosecutorial misconduct for plain error when the defendant did not object at trial, and for abuse of discretion when the district court denied an objection to closing argument. United States v. Etsitty, 130 F.3d 420, 424 (9th Cir.1997). The defendant must show that it is “more probable than not that the misconduct materially affected the verdict.” United States v. Cooper, 173 F.3d 1192, 1203 (9th Cir.1999).

During closing argument, defense counsel for Tam and Ng pointed to particular persons involved in the offense who were not called to testify, as well as to records that the prosecution had not offered into evidence. Arguably, these statements by defense counsel urged the jury to assume that the missing evidence and witness would have been favorable to the defense. The government responded to this strategy during its rebuttal argument. The prosecutor reminded the jury to consider only the evidence in front of it, but then went on to state that the defense had the same power to subpoena witnesses as did the prosecution, suggesting that if the so-called missing evidence were in fact favorable to the defendants, then the defendants would have produced it for the jury.

During a recess, defense counsel for both defendants stated to the district court that they believed the government’s argument was improper and that if made again during rebuttal that they intended to object. In response to the defense counsel, the district court stated that the government’s comments were proper, but that the repetition of them may suggest a burden on the defendant. The judge instructed the government to refrain from further comments on the subpoena power of the defense. Neither of the counsel for defendants requested a curative instruction.

We need not decide whether the government’s comments were error. Even assuming arguendo that they were, that error was rendered harmless as a result of the district court’s response to the defendants’ objection, as well as its jury instruction. The district court cured any potential error by telling the jury that the burden of proof was on the government and that the defendant need not call any witnesses or produce any evidence. Thus, even if the prosecution’s comments were improper, the defendants suffered no prejudice as a result.

Defendants Tam and Ng also assert that the district court erred in finding that the money laundering conspiracy statute does not require the indictment to allege an overt act. This was not error. The language of 18 U.S.C. § 1956 (h) is nearly identical to the language of 21 U.S.C. § 846 , which the Supreme Court held in United States v. Shabani, 513 U.S. 10, 15 , 115 S.Ct. 382 , 130 L.Ed.2d 225 (1994), does not require proof of an overt act.

Ng and Tam also argue that the district court’s jury instruction on mail fraud inadequately defined the required element of materiality so that the element was not adequately presented to the jury. This argument also fails. The instruction given by the district court was taken verbatim from the Ninth CieCuit MaNual of Model Jury InstRuctions, Instruction 8.26.1 (1997 edition). Defense counsel made no objection at trial. We note that the instruction is substantially like the definition of materiality given by the Supreme Court in United States v. Gaudin, 515 U.S. 506 , 115 S.Ct. 2810 , 132 L.Ed.2d 444 (1995). Arguably Gaudin’s definition of materiality, which includes the “capable of influencing language,” is in fact broader than the instruction given by the district court in this ease. The district court did not plainly err in its jury instruction.

Tam and Ng also both appeal their sentences, arguing that the district court erred in holding the defendants responsible for the entire scope of the money laundering scheme. In sentencing Tam and Ng on the conspiracy to commit money laundering count, the district court relied upon their respective presentence reports (“PSR”). Both of their PSRs supplied facts supporting a conclusion that the defendants knew or could have reasonably foreseen the full amount of money wire transferred and brought into the United States by Ken Mak; approximately $2,300,000. This finding resulted in a six level increase over the base offense level of 23 contained in United States Sentencing Guidelines Manual § 2S1.1 (1998).

At sentencing, the government bears the burden of proving factors enhancing a sentence by a preponderance of the evidence. United States v. Romero-Rendon, 220 F.3d 1159, 1160 (9th Cir.2000). We review de novo a district court’s compliance with Federal Rule of Criminal Procedure 32. United States v. Carter, 219 F.3d 863, 866 (9th Cir.2000). The district court’s application of the Sentencing Guidelines we review de novo. Id.

A district court’s resolution of disputed evidence related to sentencing is governed by Rule 32(c)(1), which reads as follows:

Our precedent requires “strict compliance” with Rule 32. United States v. Houston, 217 F.3d 1204, 1207 (9th Cir.2000). We adopted this rule because “resolving factual objections to the PSR on the record ensures meaningful appellate review of the sentence.” Carter, 219 F.3d at 866-67 . We hold that as to both defendants, the district court complied with Rule 32.

Prior to sentencing, both Tam and Ng objected to the six level increase recommended by their PSRs. They argued that they were responsible only for the account that they each individually controlled, or in the alternative, for the cumulative amount of the two accounts (totaling approximately $364,000, which would have resulted in a three level increase). They both disputed the facts in the PSR supporting the conclusion that they had knowledge of or could have reasonably foreseen the larger scope of the money laundering scheme.

In sentencing both defendants, the district court relied upon the defendants’ PSRs and specifically adopted the findings contained therein. In doing so, the district court indicated its awareness of the defendants’ challenges to the statements in the PSR and made the specific findings neces sitated by Rule 32. See Houston, 217 F.3d at 1209 (concluding that there was a Rule 32 violation by specifically noting that the district court had not adopted the PSR as its own findings). The facts in the defendants’ PSR are sufficient to support the district court’s finding that the government met its burden of establishing, by a preponderance of the evidence, that the defendants could have reasonably foreseen the full scope of the money laundering conspiracy.

III. As to Ng only:

Relying upon United States v. Westerdahl, 945 F.2d 1083 (9th Cir.1991), Ng raises a claim that he is entitled to an evidentiary hearing to determine whether the government intentionally distorted the fact-finding process by denying immunity to defense witnesses who might have offered relevant evidence in support of Ng’s defense. We need not decide whether this case falls within the Westerdahl exception. Even if we so found, the error in this case would be harmless.

As the government notes, the two defense witnesses who refused to testify Out of fear of prosecution were relevant only to counts one, two, and seven against Ng, and not count ten (conspiracy to launder money). In sentencing, the district court adopted the PSR’s recommendations. According to the PSR, count ten produced an offense level of twenty-nine, whereas the other counts produced an offense level of twenty-four. The PSR recommended sentencing based upon the higher level of twenty-nine. That is, Ng’s sentencing was based upon the count ten conspiracy conviction. Since the defense witnesses were not relevant to this count, their failure to testify, whether due to the government’s improper withholding of immunity or not, was harmless.

Ng also claims that it was error when, in closing argument, the prosecutor referred to what defendant Ng, who did not testify, “would say.” Claims that a prosecutor committed misconduct during summation are reviewed for plain error, when as here, trial counsel did not object. United States v. Rudberg, 122 F.3d 1199, 1206 (9th Cir.1997).

Ng points to three instances during the government’s closing argument in which the government allegedly alludes to Ng’s failure to testify. Noting that he did not testify, Ng argues on appeal that these statements are improper allusions to the defendant’s failure to testify or comments on his silence. Although defense counsel did not object at the time of trial, on appeal Ng alleges that these were Griffin violations. Griffin v. California, 380 U.S. 609 , 85 S.Ct. 1229 , 14 L.Ed.2d 106 (1965).

The test we use to determine whether there was a Griffin violation is “whether the language used was manifestly intended or was of such a character that the jury would naturally and necessarily take it to be a comment on the failure to testify.” United States v. Mende, 43 F.3d 1298, 1301 (9th Cir.1995). We have also repeatedly held that when the government refers to “defendants’ arguments” but obviously is addressing the arguments made by defense counsel, there is no Griffin violation. United States v. Sarno, 73 F.3d 1470, 1498-99 (9th Cir.1995); Mende, 43 F.3d at 1301 ; United States v. Mares, 940 F.2d 455, 461 (9th Cir.1991).

Ng’s claim that the government committed Griffin violations is without merit. As to statement (1), on its face this is a statement recounting the testimony of a witness as to what the defendant had admitted to the witness. Statements (2) and (3), which use the future tense, are comments on anticipated arguments that defense counsel will make. Furthermore, even assuming arguendo that these comments were Griffin violations, they were harmless in light of their isolated context within a long closing, their failure to directly stress the defendant’s choice not to testify, and the district court’s careful instruction as to the burden of proof remaining on the government.

On direct appeal, Ng charges that his trial counsel rendered constitutionally ineffective assistance of counsel. The record is insufficiently developed to consider these claims on direct appeal. See United States v. Daly, 974 F.2d 1215, 1218 (9th Cir.1992) (holding that ineffective assistance of counsel claims may only be raised on direct appeal when the record is sufficiently developed to permit the reviewing court to resolve the issue, or when the assistance of counsel was so blatantly ineffective that it obviously interfered with defendant’s right to counsel).

Ng also appeals his sentence, arguing that the district court erred by not departing downward on two grounds. First, he claims the court should have departed downward because his money laundering conduct was outside the “heartland” of this offense. Second, he claims that the district court erred by failing to consider a downward departure based upon disparity of sentences between him and his co-conspirators.

We lack jurisdiction to review a district court’s decision not to grant a discretionary downward departure absent evidence that the district court believed it lacked the authority to do so. United States v. Tucker, 133 F.3d 1208, 1219 (9th Cir.1998). The district court did not indicate that it believed it lacked the authority to depart on either of these bases. Therefore, we lack jurisdiction to review the district court’s discretionary refusal to depart downward.

IV. As to Giang only:

Defendant Giang argues that the evidence was insufficient to sustain his conviction. He contends that because the prosecutor’s case rested firmly upon the testimony of five accomplices with incentive to testify favorably for the government, the evidence is insufficient to com port with due process requirements. There is sufficient evidence to support a conviction if, viewing the evidence in the light most favorable to the prosecution, any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt. United States v. Coleman, 208 F.3d 786, 792-93 (9th Cir.2000).

Absent facial incredibility, it is not our role to question the jury’s assessment of witness credibility. United States v. Croft, 124 F.3d 1109, 1125 (9th Cir.1997). Furthermore, we have held that a conviction may be based on the uncorroborated testimony of a single accomplice, so long as it has not reached “a point when the witness’ qualifications are so shoddy that a verdict of acquittal should have been directed.” Lyda v. United States, 321 F.2d 788, 795 (9th Cir.1963). Defendant Giang relying on Lyda , concludes that this level of shoddiness has been reached. We disagree.

First, the jury was aware of the witnesses’ involvement in the scheme as well as their potential biases when it made its credibility finding. Their testimony is not incredible on its face. Second, independent evidence corroborated several aspects of the witnesses’ testimony. Viewing this evidence, along with the testimony of the five witnesses, in a light favorable to the prosecution, a rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt. Therefore, the evidence is sufficient to support Giang’s conviction.

Finally, Giang argues that the district court clearly erred in applying a two-point enhancement for obstruction of justice. See United States Sentencing Guidelines Manual § 3C1.1 (1998).

In order to impose' the obstruction enhancement, the district court must find (1) false testimony, (2) on a material matter, (3) with willful intent. United States v. Robinson, 63 F.3d 889, 891 (9th Cir.1995). Viewing Giang’s testimony in a light favorable to him, it does not appear that the district court clearly erred in imposing this enhancement. The record adequately reflects, as the district court found, that Giang committed perjury before the grand jury.

CONCLUSION

We AFFIRM as to all convictions and sentences of all defendants.

. Male and Kuang are fugitives. Yin, a cooperating witness, is serving an eight-year state sentence for his part in this offense. . "[T]he promises or statements were of a kind that would reasonably influence a person to part with money or property.” . Gaudin held that a false statement is material if it has "a natural tendency to influence, or [is] capable of influencing, the decision of the decisionmaking body to which it was addressed.” 515 U.S. at 509 , 115 S.Ct. 2310 . . In Westerdahl , we stated: A criminal defendant is not entitled to compel the government to grant immunity to a witness. [There is, however,] an exception to this rule in cases where tire fact-finding process is intentionally distorted by prose-cutorial misconduct.... In order to make out a claim for prosecutorial misconduct, a defendant must show that the evidence sought from the nonimmunized witness was relevant and that the government distorted the judicial fact-finding process by denying immunity to the potential witness. If a defendant makes an "unrebutted prima facie showing of prosecutorial misconduct that could have prevented a defense witness from giving relevant testimony,” we will remand the case to the district court to determine at an evidentiary hearing whether the government intentionally distorted the fact-finding process. 945 F.2d at 1086 (internal citations omitted). . The relevant passages are as follows: (1) And do you remember what [witness Philip Tham] told you about? They were giving him the run-around, and how they would do it is he would go to Robert Ng. Robert Ng would say, "Yeah. I want to talk to you about it, but Co Giang needs to be here because Co Giang recruited you in. So he is your immediate supervisor,” so to speak. "I’m the big boss. I can’t have a conversation about your performance review unless your immediate supervisor is in on the conversation,” and then he goes to Co Giang. And Co Giang says, "I've got to have Robert Ng in the conversation.” (2) Now interestingly, [Ng] tells his insurance company that he last saw the car on 6/30/94. I guess maybe he got dropped off at the airport in that car. He doesn’t report it [stolen to the police] for five days when he came back, and he is going to say that’s because of all the problems he had with his counter report maybe. (3) [One of the strawbuyers] goes, takes a couple of trips, and he gets to the dealer ship. And Ken Mak and Lewis Tam are together when he gets the money. Okay. Now the defendants are — may try and say, "Hey, once again take us out of the picture. This was all Ken Mak. These guys are wrongfully adding us into the picture.” . In Lyda , where conviction rested largely upon the testimony of one accomplice witness, we ultimately concluded: "Bearing in mind the limited role of an appellate court in a criminal appeal, we do not think that point was reached here.” 321 F.2d at 795 .