concurring:
Because I agree that the district court did not err in either calculating Perez-Alvarado’s advisory guideline range or varying upward from that range, I concur in the majority opinion. I write separately to note that, in my view, the sentence could also be affirmed, as the government urges in its supplemental brief, on the alternate reasoning given by the district court without deciding whether the court erred in its calculation of the advisory guideline range. See United States v. Keene, 470 F.3d 1347 (11th Cir.2006).1
As the majority notes, the district court used an advisory sentencing range of 63-78 months in imposing the 96-month variance sentence. To reach this range, the court used several guideline enhancements. However, the court stated that it would impose that sentence even if the sentencing range was 27-33 months, which is what Alvarado-Perez contends is appropriate:
I am going to impose a variance sentence of 96 months, and I want to state very clearly that whether the guideline recommendation produced a range of 63-78 months, which is what it did produce after I applied the guidelines, or a sentencing range of 41-51 months, which would be the way the probation officer’s original calculations had it, or a sentencing range of 27-33 months, which would be without the four level enhancement as provided by the probation officer, not any one of those sentences to me would be sufficient to reflect all of the 3553 factors here. *620So if I assume for the moment that this defendant were to prevail on every single argument on every single offense level adjustment and every single departure upward, and that I had in front of me a recommended range of sentence of 27-33 months, I would impose the exact same sentence. This is a sentence that must be imposed to reflect more accurately all of the factors set forth in Section 3553. That is a sentence that is still well below the statutory maximum in this case. It is 80 percent of a maximum sentence. And if ever were there a case where an argument could be made as has been made by the prosecutor today of seeking the most that [C]ongress would authorize, this is one of those cases. But the sentence I’ve imposed is only 80 percent of the way towards what would be recommended or authorized by [C]ongress in its wisdom in enacting this legislation. So I’m imposing a sentence of 96 months in custody-
J.A. 233-34 (emphasis added).
In Keene, the district court sentenced the defendant using a contested sentencing enhancement, but it also stated that even if the enhancement did not apply it would impose the same sentence. On appeal, the Eleventh Circuit affirmed without deciding the enhancement issue because it found that the alternate sentence was reasonable. The Eleventh Circuit termed its analysis “an assumed error harmlessness inquiry” that has two components: (1) “knowledge that the district court would have reached the same result even if it had decided the guidelines issue the other way,” and (2) “a determination that the sentence would be reasonable even if the guidelines issue had been decided in the defendant’s favor----” 470 F.3d at 1349. Applying that analysis to the facts before it, the Eleventh Circuit concluded that even if there was any misapplication of the guideline enhancement,
“the error did not affect the district court’s selection of the sentence imposed.” Put a little differently, it would make no sense to set aside this reasonable sentence and send the ease back to the district court since it has already told us that it would impose exactly the same sentence, a sentence we would be compelled to affirm.
Id. at 1350 (quoting Williams v. United States, 503 U.S. 193, 203, 112 S.Ct. 1112, 117 L.Ed.2d 341 (1992)). See also United States v. Dean, 517 F.3d 1224, 1232 (11th Cir.2008), cert. denied sub nom. Lopez v. United States, — U.S. -, 129 S.Ct. 2155, 173 L.Ed.2d 1158 (2009) (applying the Keene analysis to affirm the variance sentence).
Improper calculation of an advisory guideline range constitutes a procedural error, Gall v. United States, 552 U.S. 38, 51, 128 S.Ct. 586, 169 L.Ed.2d 445 (2007), and “procedural errors at sentencing ... are routinely subject to harmlessness review,” Puckett v. United States, — U.S. -, 129 S.Ct. 1423, 1432, 173 L.Ed.2d 266 (2009). Thus, “[i]n evaluating a district court’s error in its sentencing calculations, we must determine whether the error was harmless.” United States v. Mehta, 594 F.3d 277, 283 (4th Cir.2010). A sentencing calculation error is harmless “if the resulting sentence was not ‘longer than that to which [the defendant] would otherwise be subject.’ ” Id. (quoting United States v. Stokes, 261 F.3d 496, 499 (4th Cir.2001)).
I am not aware of any published decision in which we have applied the Keene analysis in the precise situation before us: that is, where we have entirely given the benefit of the doubt regarding the appropriate advisory sentencing range to the defendant and then simply decided the reasonableness of an alternate variance sentence *621from that range.2 However, such an approach seems entirely consistent with our caselaw.
For example, it is not uncommon for us to assume, without deciding, error in the context of conducting a harmless error analysis. See, e.g., United States v. Johnson, 587 F.3d 625, 637 (4th Cir.2009), cert. denied sub nom. Martin v. United States, -U.S.-, 130 S.Ct. 2120, 176 L.Ed.2d 744, 78 U.S.L.W. 3611 (2010) (“Martin raises several potential errors that the district court made in admitting the 1980 conviction, but we need not address them. Assuming that the district court erred in admitting the conviction, that error was harmless.”); United States v. Lipford, 203 F.3d 259, 273 (4th Cir.2000) (“[W]e need not decide whether the conduct here meets that broad definition because even if we assume, without deciding, that there was error, any such error would be harmless.”).
Indeed, we recently did so in a sentencing case. In United States v. Boulware, 604 F.3d 832 (4th Cir.2010), the defendant challenged her sentence on appeal, arguing that the district court did not offer sufficient reasons to show that it made an individual assessment of the specific circumstances in her case in light of the relevant § 3553(a) factors. Without deciding whether such an error occurred, we rejected this argument because the alleged error was harmless:
In light of the strong indications that the district court fully considered Boulware’s argument for a below-guidelines sentence, and in light of the weakness of that argument, the notion that having to explain its analysis further might have changed the district court’s mind ... is simply unrealistic in the present case, and remand for resentencing would be a pointless waste of resources. We therefore hold that even assuming that the district court committed procedural error in failing to sufficiently explain the sentence imposed in light of the § 3553(a) factors, any error was harmless.
604 F.3d 832, 840.
Moreover, in a slightly different context, we have affirmed a sentence that was erroneously imposed because we concluded that the district court’s statement of an alternate sentence rendered the error harmless. See United States v. Shatley, 448 F.3d 264 (4th Cir.2006). We explained in that case:
Thus, even though the district court did impose a sentence under the Sentencing Guidelines that violated Shatley’s Sixth Amendment rights, it announced at the same time that if the Sentencing Guidelines became advisory, it would impose the same sentence. In making that determination, it properly considered the § 3553(a) factors____ We take the district court at its word when it stated plainly that it would impose the same sentence under an advisory guideline system. Thus, the error was harmless; to remand this case now in view of these facts would amount to an “empty formality.”
448 F.3d at 268 (citation omitted).
We have come close to applying the Keene analysis on at least two occasions. For example, in United States v. Evans, 526 F.3d 155 (4th Cir.), cert. denied, — U.S.-, 129 S.Ct. 476, 172 L.Ed.2d 341 (2008), the defendant did not challenge the calculation of his advisory sentencing range, but he did challenge the district *622court’s use of a guideline provision to depart upward from the range. Without deciding whether the court erred in its use of the guideline provision, we found the ultimate sentence to be reasonable because the court imposed it by relying on both the guideline provision and, independently, the § 3553(a) factors. We noted: ‘When, as here, a district court offers two or more independent rationales for its deviation, an appellate court cannot hold the sentence unreasonable if the appellate court finds fault with just one of these rationales.” 526 F.3d at 165 (emphasis in original). We then concluded: “Accordingly, even assuming the district court erred in applying the Guideline departure provisions, Evans’ sentence, which is well-justified by § 3553(a) factors, is reasonable.” Id.
More recently, in United States v. Grubbs, 585 F.3d 793 (4th Cir.2009), cert. denied, — U.S.-, 130 S.Ct. 1923, 176 L.Ed.2d 392 (2010), the defendant argued that the district court committed two procedural errors involving two enhancements used to calculate his guideline range. We rejected the defendant’s argument as to the first enhancement, finding it to be proper. However, we noted regarding the second enhancement that the court “may have erred.” 585 F.3d at 804. Nonetheless, we affirmed the sentence, finding it to be procedurally reasonable “because the district court adequately explained its sentence on alternative grounds supporting a variance sentence, by reference to the 18 U.S.C. § 3553(a) factors.” Id.
As the foregoing authorities illustrate, (1) we review procedural sentencing errors (such as guideline range calculations) for harmlessness, (2) we have previously assumed without deciding that sentencing errors occurred in affirming sentences under a harmless error analysis, and (3) we have accepted a district court’s statement of an alternate sentence as a basis to conclude that a sentencing error was harmless. Taken together, these analytical approaches form the Keene “assumed error harmlessness” analysis.
Applying that analysis to this case, we (1) initially give Alvarado-Perez the benefit of the doubt and assume (without deciding) that he is correct that the sentencing range is 27-33 months and (2) proceed to decide the question of whether a variance from that range to 96 months is reasonable. Based on the reasons stated by the district court, and in light of our deferential standard of review, I believe that the variance from 27-33 months to 96 months is reasonable. Accordingly, the sentence could be affirmed on this basis without reaching the merits of the claimed guideline calculation errors.
. Alternate sentences are not completely uncommon. Indeed, in a more limited context, we have previously recommended that district courts impose alternate sentences. See United States v. Hammoud, 378 F.3d 426 (4th Cir.2004) (en banc order holding that the sentencing guidelines are mandatory but recommending that district courts impose alternate sentences treating the guidelines as advisory pending Supreme Court guidance), vacated, 543 U.S. 1097, 125 S.Ct. 1051, 160 L.Ed.2d 997 (2005) (for further consideration in light of United States v. Booker, 543 U.S. 220, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005)).
. Cf. United States v. Lee, 321 Fed.Appx. 298, 302 (4th Cir.) (Shedd, J., concurring) (noting that the sentence could have been affirmed under the Keene analysis), cert. denied, - U.S. -, 130 S.Ct. 204, 175 L.Ed.2d 142 (2009).