The government appeals, contending that the defendant’s non-Guidelines sentence, which was substantially below the advisory range under the Guidelines, is unreasonable. The question presented is whether the sentence is unreasonable because the district court departed downward based on the conclusion that imposing a Guidelines sentence would create an unwarranted disparity between the defendant’s sentence and the national average *543for sentences under the same Guidelines section. We hold that the sentence is unreasonable and remand for resentencing.
I.
Douglas Willingham used accounts at the Yahoo! and Photoisland internet sites to view, store and trade images of child pornography online. When law enforcement officers executed a search. warrant for his office computer on October 21, 2004, they found 197 images depicting graphic child pornography, including images of extremely young children.
Willingham was charged with one count of possession of child pornography in violation of 18 U.S.C. § 2252(a)(4)(B). Pursuant to a plea agreement and stipulated facts, he pled guilty. The U.S. Probation Office prepared a pre-sentence report (“PSR”), in which it determined Willing-ham’s final offense level to be 30. This offense level was calculated using a base level of 17 points and included enhancements for: involvement of minors younger than 12 ( + 2); distribution of images in exchange for value ( + 5); portrayal of sadistic conduct ( + 4); use of a computer (+2); and possession of between 150 and 300 images (+3). The PSR also deducted three points for Willingham’s acceptance of responsibility. The advisory sentencing range for an offense level of 30 under U.S.S.G. § 2G2.2 was 97-120 months.1
Willingham filed an objection to the PSR, arguing that a sentence within the advisory range would constitute an “unwarranted disparity” between his sentence and those of other offenders sentenced under § 2G2.2. His argument was based on statistics from the U.S. Sentencing Commission’s March 2006 “Final Report on the Impact of United States v. Booker on Federal Sentencing”2 (“the Report”). The Report included a table with average and median sentences for criminals sentenced under § 2G2.2 between October 2002 and April 2003 and May 2003 and June 2004. The first period showed an average term of 65 months and median sentence of 51 months. During the second period the average sentence was 63 months and the median was again 51 months. Based on this evidence, Willing-ham argued that to avoid a disparity in sentencing, he too should be sentenced to a term of approximately 60 months. •
At a sentencing hearing held on September 14, 2006, the district court-briefly mentioned the sentencing factors enumerated in 18 U.S.C. § 3553(a) before emphasizing the importance of Willingham’s statistical evidence on its decision to impose a sentence of 64 months. See infra note 3. The government merely noted for the record that it objected to the non-Guidelines sentence, but it did not offer' any argument against Willingham’s statistical evidence, nor did it specify grounds for .the objection. The government filed this appeal seeking vacatur of the 64-month sentence and a remand for resentencing.
II.
We generally review a non-Guidelines sentence for unreasonableness. United States v. Booker, 543 U.S. 220, 261, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005). A non-Guidelines sentence “unreasonably fails to reflect the statutory sentencing factors where it (1) does not account for a factor that should have received significant weight, (2) gives significant weight to an *544irrelevant or improper factor, or (3) represents a clear error of judgment in balancing the sentencing factors.” United States v. Smith, 440 F.3d 704, 708 (5th Cir.2006).
Complicating our review somewhat, the government admits that it is making arguments on appeal that it did not present to the district court. Thus we will review the district court’s sentencing decision under the “plain error” standard and ask whether the government can show “(1) a plain (‘clear’ or ‘obvious’) error that (2) affected [its] substantial rights.” United States v. Rodarte-Vasquez, 488 F.3d 316, 321 (5th Cir.2007) (citing United States v. Olano, 507 U.S. 725, 732-35, 113 S.Ct. 1770, 123 L.Ed.2d 508 (1993)). If the government makes this showing, we “may then exercise [our] discretion to notice a forfeited error but only if ... the error seriously affects the fairness, integrity, or public reputation of judicial proceedings.” Rodarte-Vasquez, 488 F.3d at 321 (citing United States v. Mares, 402 F.3d 511, 520 (5th Cir.2005)).
We first turn to decide whether the alleged error of the district court was plain. Once a district court has calculated the correct guidelines range, it must use that range as a “frame of reference” and then weigh the factors listed in § 3553(a). Smith, 440 F.3d at 707. A district court’s reasons for a downward departure must be both thoroughly articulated and include facts specific to the individual defendant. See Mares, 402 F.3d at 519. Three § 3553(a) factors are at least partially ease-specific, see United States v. Williams, 456 F.3d 1353, 1369 (11th Cir.2006):
(1) the nature and circumstances of the offense and the history and characteristics of the defendant;
(2) the need for the sentence imposed- —
(A) to reflect the seriousness of the offense, to promote respect for the law, and to provide just punishment for the offense;
(B) to afford adequate deterrence to criminal conduct;
(C) to protect the public from further crimes of the defendant; and
(D) to provide the defendant with needed educational or vocational training, medical care, or other correctional treatment in the most effective manner .... [and]
(6) the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct[.]
§ 3553(a) (emphasis supplied). Relevant to this appeal, the case-specific factor applied by the district court was (a)(6), as Willingham argued below that imposing a Guidelines sentence would have resulted in an improper sentence disparity between him and the average offender.
When we consider the district court’s application of § 3553(a)(6) in imposing this non-Guidelines sentence, we think that the district court committed clear error. Although the statistics may show a disparity between the average § 2G2.2 sentence and the advisory Guidelines range in Willingham’s case, there is no indication that the disparity is unwarranted. National averages of sentences that provide no details underlying the sentences are unreliable to determine unwarranted disparity because they do not reflect the enhancements or adjustments for the aggravating or mitigating factors that distinguish individual cases. Here, the pre-sentence report indicated that 13 net points were added to Willingham’s offense level under the Guidelines, reflecting several serious aggravating factors. The averages in the Report disregard individual circumstances and only reflect a broad *545grouping of sentences imposed on a broad grouping of criminal defendants; consequently, they are basically meaningless in considering whether a disparity with respect to a particular defendant is warranted or unwarranted.
Clear error is further indicated by the language of § 3553(a)(6), which makes plain that an unwarranted disparity is to be avoided between defendants with “similar records who have been found guilty of similar conduct.” Here, it is apparent that the statistics in the Report tell us nothing about the similarity (or lack thereof) of Willingham’s record or his conduct when compared to those sentenced under § 2G2.2 that collectively compose the national averages. Indeed, we agree with the Seventh Circuit, when it recently observed that a reviewing court’s concern about unwarranted disparities is at a minimum when a sentence is within the Guidelines range. See United States v. Boscarino, 437 F.3d 634, 638 (7th Cir.2006). In sum, the error of the district court was plain.
Next we turn to the second part of the plain error analysis, that is, whether the government can show that its substantial rights were affected. The answer depends upon whether, absent the error of misapplying § 3553(a)(6), the sentence imposed on Willingham likely would have been the same. We thus look to the record of the sentencing hearing to discern whether the district court pointed to other factors or reasoning supporting the 64-month sentence imposed.
A review of the transcript of the sentencing proceedings, reproduced below,3 *546makes clear that the district court’s non-Guidelines sentence was imposed in response to Willingham’s argument based on the national statistics and not on any other individual factor under § 3553(a); thus, there is no indication that, absent the error, the sentence would have been the same. Accordingly, we find that this plain error affected the government’s substantial right to have a proper and lawful sentence imposed on the defendant.
The third and final step in plain error review is to consider whether we should exercise our discretion to correct this error because the error “seriously affects the fairness, integrity, or public reputation of judicial proceedings.” Rodarte-Vasquez, 488 F.3d at 321. In effect, this step brings us to the question of whether Willingham’s sentence was reasonable, which of course is the standard of review for all non-Guideline sentences. See Booker, 543 U.S. at 261, 125 S.Ct. 738. Based on our holding in Smith, the sentence is unreasonable because it gave significant weight to an irrelevant or improper factor, which was the Commission’s reports.
III.
For the foregoing reasons, Willingham’s sentence is VACATED and the case is REMANDED for resentencing.
VACATED and REMANDED.