United States v. Precias Freeman

PUBLISHED

UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT

No. 19-4104

UNITED STATES OF AMERICA,

Plaintiff – Appellee,

v.

PRECIAS K. FREEMAN,

Defendant – Appellant.

Appeal from the United States District Court for the District of South Carolina, at Spartanburg. Timothy M. Cain, District Judge. (7:17-cr-00079-TMC-1)

Argued: October 30, 2020 Decided: March 30, 2021 Before GREGORY, Chief Judge, FLOYD, and QUATTLEBAUM, Circuit Judges.

Vacated and remanded by published opinion. Chief Judge Gregory wrote the opinion, in which Judge Floyd joined. Judge Quattlebaum wrote a dissenting opinion.

ARGUED: Hannah Rogers Metcalfe, METCALFE & ATKINSON, LLC, Greenville, South Carolina, for Appellant. William Jacob Watkins, OFFICE OF THE UNITED STATES ATTORNEY, Greenville, South Carolina, for Appellee. ON BRIEF: Peter M. McCoy, Jr., United States Attorney, OFFICE OF THE UNITED STATES ATTORNEY, Columbia, South Carolina, for Appellee.

GREGORY, Chief Judge:

Precias Freeman broke her tailbone as a teenager, was prescribed opioids, and has been addicted to the drugs ever since. In 2018, she was sentenced to serve more than 17 years in prison for possession with intent to distribute hydrocodone and oxycodone in violation of 21 U.S.C. §§ 841(a)(1) and 841(b)(1)(C). After Freeman’s appointed counsel initially submitted an Anders brief asking for the Court’s assistance in identifying any appealable issues, we directed counsel to brief whether Freeman’s sentence is substantively reasonable and whether Freeman received ineffective assistance of counsel on the face of the record. On both grounds, we vacate Freeman’s sentence and remand this case for resentencing.

I.

Freeman pleaded guilty without the benefit of a plea agreement to an indictment charging her with possession with intent to distribute hydrocodone and oxycodone. 21 U.S.C. §§ 841(a)(1) and 841(b)(1)(C). She was charged and sentenced for conduct occurring between October 2014 and October 2016. But as reflected in her criminal history and according to statements she made to the government and the court, Freeman’s opioid addiction and pattern of filling forged prescriptions in order to obtain opioids began in 2000, when she was about 18 years old. During her years of addiction and criminal activity, there is no indication that Freeman was ever violent or associated with anyone engaged in violence. Most of the pills that she sold, including all of those sold between 2014 and *3 2016, were sold below market rate to the same woman. At the time of her arrest, Freeman was in considerable debt.

Freeman was first prescribed opioids as a teenager after breaking her tailbone in the shower. In the most comprehensive interview regarding her conduct, Freeman told the government that the doctor for whom she worked at the time permitted her to write her own prescriptions for the pain medication Lortab, or hydrocodone, beginning with 30-pill prescriptions containing 5 milligrams of hydrocodone each. 1 “[E]ver since then,” she told the government, she has been “hooked” on hydrocodone. Around 2001, while working at another medical practice and while still a teenager, Freeman started printing duplicate prescriptions for patients prescribed opioids and keeping one for herself. Once she filled these duplicate prescriptions, she would use half of the pills and sell the other half to an acquaintance who worked in a hospital as a lab technician. She eventually began writing forged prescriptions.

Over time, Freeman’s fraudulent prescriptions contained more and more pills at higher and higher concentrations, with the amount of prescriptions she filled varying with her personal use of the drugs. By October 2014, the beginning of the period relevant to Freeman’s federal charge, Freeman told federal investigators that she was filling “one prescription per day, four to five days per week.” She used some of the pills and sold

1 These facts principally emerge from a proffer interview memorialized by a Drug Enforcement Agency officer that was, according to the parties and the district court, meant to be the basis for Freeman’s accountability at sentencing.

others. By February 2015, her own use had increased to 60 to 80 tablets per day—more than half of the total pills from the forged prescriptions that she was filling.

In 2008 and 2011, Freeman’s conduct resulted in state convictions for obtaining fraudulent prescriptions and related crimes. Her criminal record also shows similar state charges that the state declined to prosecute. All of Freeman’s prior conduct relates to using and selling opioids. Relevant to this appeal, Freeman was eventually arrested on state charges on October 2, 2016, after a Walgreen’s pharmacist recognized her and called police. She was then transported to a hospital, where she tested positive for opiates. That same day, state investigators went to interview Freeman at the hospital. She spoke to them after waiving her Miranda rights. While Freeman was incarcerated on the pending state charges, a federal grand jury returned an indictment charging her with possession with intent to distribute hydrocodone and oxycodone. §§ 841(a)(1) and 841(b)(1)(C). 2

While awaiting sentencing, Freeman spoke to the government pursuant to a standard proffer agreement. During this interview, Freeman conservatively estimated that she sold 52,000 10-mg tablets of hydrocodone to her drug buyer between October 2014 and October 2016. No agreement emerged from Freeman’s proffer. Instead, while she was awaiting sentencing and released on bond, Freeman left South Carolina with her family in September 2017. Shortly before she left, Freeman failed an instant drug test and admitted

2 The indictment also states that Freeman “intentionally did combine, conspire, agree and have tacit understanding” with others to distribute hydrocodone and oxycodone, and cites that her conduct thus also violated 21 U.S.C. § 846 (Attempt and Conspiracy). But § 846 does not expose Freeman to any additional liability in this case, because the government has only held Freeman accountable for conduct that she herself admittedly committed in violation of §§ 841(a)(1) and 841(b)(1)(C).

she had taken Lortab. On the basis of the test, her probation officer sought to modify Freeman’s bond to require GPS monitoring, which was ordered by the court on September 7, 2017. However, it does not appear from the record that Freeman left the jurisdiction due to this change in her probation. During this same time period, public records confirm that Freeman and her family were evicted from their apartment, and on September 7th or 8th began living in hotels near their hometown of Shelby, North Carolina, about 40 miles away from their former home in South Carolina. 3 As Freeman explained to the district court at her sentencing, the family—including four children and a pregnant Freeman—left because they “didn’t have anywhere to go.” Freeman was rearrested in March 2018. Between September and March, Freeman remained in and around Shelby with her family. Freeman also gave birth during this time. The docket does not reflect that Freeman missed any court dates or ever attempted to evade arrest between September 2017 and March 2018.

In July 2018, a few months after she was rearrested, Freeman appeared before the district court for a sentencing hearing. The government presented evidence that she had obtained 59 fraudulent prescriptions, each between 90 and 120 pills, including evidence that she filled five prescriptions on December 1, 2014, and 13 prescriptions on December

3 In the presentence report (PSR), the court probation officer notes Freeman’s address and states that she was living with her parents at the time that the PSR was first prepared. The docket number of the family’s eviction case appears in the record and corresponds to a public eviction case under the name of “Freeman” and relating to the same address and apartment number listed as Freeman’s in the PSR. The eviction was final on September 8, 2017. The government does not argue that the eviction did not take place or that the court records are inaccurate. This Court takes judicial notice of the fact of the eviction, which is a matter of public record. See Massachusetts v. Westcott, 431 U.S. 322, 323 n.2 (1977); Hall v. Virginia, 385 F.3d 421, 424 n.3 (4th Cir. 2004).

30, 2014. But the probation officer determined that Freeman was responsible for filling far more than just those prescriptions. Based on Freeman’s initial statement to state police, given while she was hospitalized and positive for opioids, the probation officer estimated that Freeman had successfully filled one prescription per day every day for two years, 365 days a year. She accordingly held Freeman responsible for obtaining with intent to distribute 87,600 tablets of hydrocodone—the equivalent, for purposes of sentencing, of 5,869.2 kilograms of marijuana. The presentence report (PSR) did not reduce this number to reflect Freeman’s own use of the pills, which could not be the basis for a charge of possession with intent to distribute. The final calculated drug weight corresponded to an offense level of 32. Overall, Freeman’s calculated offense level was 34.

At the hearing, Freeman raised questions about the drug weight assessed in the PSR. She informed the district court that she was having a hard time contacting her counsel and that she disagreed with her counsel about how best to proceed with her case. In response, the district court continued the hearing, and Freeman’s family hired another attorney to represent her. The district court noted that the government would in the meantime revisit the drug weight amount based on the information in Freeman’s proffer, noting that the drug weight could rise or fall accordingly, and the government agreed. The agreement to revise the PSR on the basis of the proffer was even memorialized on the docket.

The probation officer did revise the PSR, but not based on Freeman’s proffer. In the new report, the probation officer “conservative[ly]” estimated that Freeman successfully obtained and intended to distribute two prescriptions of 120 10-mg pills every day of the week for two years, 365 days per year, again with no reduction for Freeman’s *7 own significant personal use of the pills. That amount of drugs—175,200 pills—is the equivalent, for purposes of sentencing, of 11,738.4 kilograms of marijuana. The probation officer accordingly assigned Freeman a base offense level of 34. This amount was significantly higher than the 52,000-pill estimate that Freeman had conservatively estimated she had sold in her proffer. The probation officer also assigned Freeman a two-level increase in offense level for obstruction of justice based on her moving from the jurisdiction, and did not recommend a three-level decrease in offense level that Freeman otherwise would have received for accepting responsibility by admitting her conduct and cooperating with law enforcement. Freeman’s recalculated overall offense level was 36.

At Freeman’s rescheduled sentencing hearing, the government and the defense both stated (incorrectly) that a review of the proffer had led to the revised PSR’s increase in Freeman’s assigned drug weight. Prior to the hearing, Freeman’s new attorney had lodged objections to the PSR related to Freeman’s failed drug test; the government’s lack of evidence for its calculated drug weight; and the facts relating to obstruction of justice. But on the day of the hearing he waived these objections, apparently to Freeman’s surprise and counter to their agreed-upon strategy. Freeman’s counsel told the district court that he was waiving the objections because they “might be considered as minimal”; were “not going to change, in essence, what the charges are”; and (incorrectly) would not “reduce the number that is relevant to this Court.” 4 When the district court asked Freeman if she was OK with

4 In a post-sentencing filing, counsel also attributed the decision to waive the objections to various factors, some of them incomprehensible. These included the government’s “inability to prosecute for several years. Agent versus Defendant in a matter (Continued)

waiving her objections, she initially responded, “I’m not sure.” After talking to her counsel, Freeman agreed to the waiver.

Instead of pursuing the objections, Freeman’s counsel relied entirely on a motion to enter a drug court diversion program (the “BRIDGE program”) that could have permitted Freeman to enter treatment instead of going to prison. Emails in the record suggest that counsel did not understand how to obtain entry into the program or what the district court would need in order to grant a motion to enter it; he wrote in an email to the program’s supervising probation officer that he was “not completely aware of the parameters of the drug court.” Nothing in the record suggests that Freeman was ever screened for drug court by probation; indeed, as counsel wrote in the motion requesting that she be considered for the program, “[d]ue to the significant sentence in front of her, it is thought that the BRIDGE [p]rogram should not be presented to her.” Counsel did append evidence that Freeman had been diagnosed with severe opioid use disorder. Freeman also spoke at the hearing. She apologized to her family and to the court for her actions, and asked the court for mercy.

The district court denied the motion for the BRIDGE program and sentenced Freeman to 210 months, the low end of the Guidelines calculation based on the offense level in the PSR. Freeman’s counsel then filed, “pursuant to Federal Rule of Civil Procedure Rule 59(e),” a petition for rehearing or motion for reconsideration. Noting that the Federal Rules of Civil Procedure do not apply in criminal proceedings, the district court

of credibility[]” and Freeman’s “quintessential case for the Bridge [drug diversion] program.” In the filing, Freeman’s attorney did not seem to understand the effect of his waiver, continuing to argue points that he had waived and contending that he had “specifically provided the Court with objective evidence of [Freeman’s] objections.”

construed the motion as one to modify the sentence under Federal Rule of Criminal Procedure Rule 35. Finding that Freeman’s counsel had not informed the court of any “arithmetical, technical, or other clear error” within the ambit of Rule 35, the district court denied the motion.

II.

Criminal defendants are entitled to effective assistance from counsel. U.S. CONST. amend. VI. This right extends to sentencing proceedings. Lafler v. Cooper, 566 U.S. 156, 165 (2012). A lawyer is constitutionally ineffective where her representation falls below objective standards of reasonableness and results in prejudice, meaning there is “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Strickland v. Washington, 466 U.S. 668, 687–88, 694 (1984). Reviewing courts are “highly deferential” to the strategic decisions of counsel, which should be judged on the facts of a given case and from counsel’s perspective at the time that the decisions were made. Id. at 689–90. But a decision cannot be considered “tactical” where “it made no sense or was unreasonable.” Vinson v. True, 436 F.3d 412, 419 (4th Cir. 2006).

Freeman’s ineffective assistance of counsel claim is made on direct appeal and therefore was not litigated before the trial court. We review it de novo, but will reverse only if it “conclusively appears in the trial record itself” that the defendant did not receive effective representation. United States v. Fisher, 477 F.2d 300, 302 (4th Cir. 1973) (quoting United States v. Mandello, 426 F.2d 1021, 1023 (4th Cir. 1970)).

To prevail on such a claim, a defendant must “demonstrate that counsel’s performance was deficient with respect to prevailing professional norms or duties.” United States v. Carthorne, 878 F.3d 458, 466 (4th Cir. 2017). These include “the duty to investigate and to research a client’s case in a manner sufficient to support informed legal judgments.” Id. Counsel is “quintessential[ly]” deficient when he does not know the law that is fundamental to his case and fails to conduct research on highly relevant points of law. Id. (quoting Hinton v. Alabama, 571 U.S. 263, 274 (2014)). “An attorney’s failure to object to an error in the court’s guidelines calculation that results in a longer sentence for the defendant can demonstrate constitutionally ineffective performance.” 878 F.3d at 467 (quoting Ramirez v. United States, 799 F.3d 845, 855 (7th Cir. 2015)). An attorney’s decision to affirmatively waive meritorious objections can likewise constitute ineffective performance.

In this case, Freeman’s attorney failed to lodge a meritorious objection to the PSR’s calculated drug weight. The docket and a court transcript indicate that the district court and the government agreed that Freeman’s PSR would be revised based on her proffer. Revision on that basis should have resulted in a base offense level of 32. No matter how the proffer was taken into account—even if the government were to double Freeman’s conservative estimate of the number of pills she sold between October 2014 and October 2016, raising her drug weight to the equivalent of 102,000 pills—an objection to the assessed drug weight on this basis would have resulted in a lower Guidelines range, from 210–262 months to 168–210 months. See U.S.S.G. ch. 5, pt. A (Sentencing Table). Yet *11 upon revising the PSR, the probation officer nonetheless increased the drug weight to an amount corresponding to a base offense level of 34.

Freeman’s counsel also waived an objection related to a two-level upward adjustment for obstruction of justice based on Freeman leaving South Carolina while on bond. This upward adjustment had another effect: the probation officer used it to disqualify Freeman from an available three-level downward departure for acceptance of responsibility. A successful objection to the facts underlying the obstruction of justice departure could accordingly have resulted in a total offense level that was five levels lower than the one Freeman was ultimately assigned.

Relevant to this case, the Guidelines explain that an upward departure for obstruction of justice is appropriate where (and only where) “the defendant willfully obstructed or impeded, or attempted to obstruct or impede, the administration of justice with respect to the investigation, prosecution, or sentencing of the instant offense of conviction, and . . . the obstructive conduct related to . . . the defendant’s offense of conviction and any relevant conduct[.]” U.S.S.G. § 3C1.1. The application notes state that the enhancement may be appropriate where the conduct involves “escaping or attempting to escape from custody before trial or sentencing; or willfully failing to appear, as ordered, for a judicial proceeding[.]” Id. at Application Note 4(E). But the notes go on to state that the enhancement is ordinarily not warranted for “avoiding or fleeing from arrest.” Id. at Application Note 5(D). In other words, “[m]erely avoiding or fleeing from arrest does not warrant an adjustment.” United States v. Jordan, 100 F.3d 950 (Table) (4th Cir. 1996); see also United States v. Lara, 472 Fed. App’x 247, 248–49 (4th Cir. 2012) (same).

Freeman no doubt violated her bond by leaving South Carolina for North Carolina after being evicted. But that does not necessarily mean that she “escape[d] or attempt[ed] escape from custody,” or otherwise willfully obstructed or attempted to obstruct justice. See U.S.S.G. § 3C1.1. This Court has held that an obstruction enhancement for flight is appropriate in a case where a defendant fled; eluded officers; obtained and began using fake identification; failed to appear at sentencing; and told officers that he “had no intention of turning himself in and . . . would do whatever was needed to remain at large until he was located and arrested.” United States v. Hudson, 272 F.3d 260, 262–64 (4th Cir. 2001). It has similarly held that the enhancement was appropriate when a defendant fled to Florida from North Carolina, told his probation officer he was going to South America, and missed his sentencing hearing. United States v. Miller, 77 F.3d 71, 73–74 (4th Cir. 1996); see also United States v. Brown, 438 Fed. App’x 203, 204 (4th Cir. 2011) (finding enhancement appropriate where the defendant missed a hearing and was a fugitive for more than eight years). In contrast to those cases, Freeman left after being evicted; was pregnant; went to her hometown just 40 miles away with her family and children in tow; and called her probation officer to tell her what happened. She did not even miss a court date.

A reading of the relevant Guidelines and Application Notes alone should have indicated to Freeman’s counsel that an objection to the adjustment was available and could result in significantly less sentencing exposure for Freeman. But counsel waived his client’s objection to the obstruction enhancement because, he told the court, it would not affect his client’s sentence. He was incorrect on the law.

Even if the obstruction of justice enhancement were found to apply, however, counsel erred by failing to object to the PSR’s failure to apply the downward adjustment for acceptance of responsibility. As the application notes expressly contemplate, it is possible to receive both an upward departure for obstruction of justice and a downward departure for acceptance of responsibility in “extraordinary” cases. See U.S.S.G. § 3E1.1 Application Note 4. In this Circuit, whether a case is “extraordinary” in this way is “a largely factual matter to be determined by the district court.” United States v. Knight, 606 F.3d 171, 177 (4th Cir. 2010). It is not foreclosed in any given case. This Court has affirmed such a Guidelines calculation in an “extraordinary case” where, despite a defendant’s flight from officers and lies to a probation officer, he also cooperated significantly. United States v. Hicks, 948 F.2d 877, 879–81, 885 (4th Cir. 1991). Freeman significantly cooperated with the government; was severely addicted to opioids; had been told that the Department of Social Services would attend her next meeting with her probation officer and that her children might be taken from her; and was pregnant at the time she was forced to leave her home in South Carolina, with no place for her and her children to live. The sentencing court may have appropriately concluded that Freeman’s was the “extraordinary case” where a defendant still merited a downward departure for acceptance of responsibility, resulting in a Guidelines range of 151–188 months absent any other reductions. But counsel did not argue the point.

Counsel variously states in the record that he waived his client’s objections to the PSR because they “might be considered as minimal”; were “not going to change, in essence, what the charges are”; and would not “reduce the number that is relevant to this *14 [c]ourt.” Later contradicting himself in an improper post-sentencing filing, he continued to argue the objections while variously stating that he waived them because the court was aware of them; because the government was not credible; and because his client was a “quintessential case for the [drug diversion] program.” But as this Court explained in United States v. Richardson, an express, “knowing[,] and voluntary” waiver at sentencing is almost always binding. 744 F.3d 293, 299–300 (4th Cir. 2014). Counsel’s attempt to resurrect these issues immediately after waiving them indicates that he may have improperly advised his client regarding waiver at the sentencing hearing. Separately, Freeman stated in a pro se filing to this Court that her counsel advised her to waive the objections because arguing them would create a “hostile environment” in the courtroom. That may be an appropriate justification for foregoing a meritless objection, but it would never be an appropriate justification for waiving a meritorious one.

The multiple justifications available on the face of the record suggest that counsel went into his client’s sentencing woefully unprepared. 5 But even considered individually, none would be reasonable. Most notably, counsel was wrong that the objections would not affect his client’s sentence. With a two-level reduction for drug weight; a two-level reduction for not receiving the obstruction of justice enhancement; and a three-level reduction for acceptance of responsibility in the absence of the obstruction enhancement,

5 The numerous, inconsistent, and shifting justifications for counsel’s decision to waive the objections—already available on the face of the record—also confirm our determination that waiting to address Freeman’s ineffective assistance claim on collateral review would not provide this Court with information useful to adjudicating this claim.

Freeman could appropriately have been sentenced at an offense level of 29, with a Guidelines range of 97 to 121 months. 6 Unlike our dissenting colleague, we are confident on this record that counsel had no strategic reason to waive straightforward legal arguments that could result in his client receiving approximately a decade less time behind bars in favor of a nonexclusive motion for entry into a diversion program.

On this record, it appears that counsel did not understand his client’s sentencing exposure nor the law fundamental to his client’s objections. His performance in this case was thus the kind of quintessentially ineffective assistance that lies beneath the constitutional floor. See Carthorne, 878 F.3d at 466. However, for this Court to vacate and remand Freeman’s case for resentencing on her ineffective assistance claim, her counsel’s performance must also have prejudiced her. Strickland, 466 U.S. at 687–88.

In this case, the prejudice is manifest. A defendant establishes prejudice under Strickland where there is a “‘reasonable probability’ that the outcome of a sentencing would change” absent counsel’s errors. Carthorne, 878 F.3d at 470 (quoting United States v. Rangel, 781 F.3d 736, 746 (4th Cir. 2015)). “[I]n most cases, when a district court adopts

6 At sentencing, the district court stated that it would have sentenced Freeman to 210 months imprisonment regardless of the Guideline range. While a district court may depart from the Guideline range in a particular case, the Guidelines have a “central role” in sentencing and a “real and pervasive effect” on sentence length. Molina-Martinez v. United States, 136 S.Ct. 1338, 1345–46 (2016). Guideline ranges “serve as the starting point for the district court’s decision and anchor the court’s discretion in selecting an appropriate sentence.” Id. at 1349. When the calculated Guideline range is incorrect, “a defendant . . . should be able to rely on that fact to show a reasonable probability that the district court would have imposed a different sentence under the correct range. That probability is all that is needed to establish an effect on substantial rights[.]” Id. Under the circumstances of this case, as explained below, it is likewise all that is needed to establish prejudice under Strickland.

an incorrect Guidelines range, there is a reasonable probability that the defendant’s sentence would be different absent the error.” Molina-Martinez v. United States, 136 S. Ct. 1338, 1341 (2016). Here, the incorrect drug weight calculation alone would have entitled Freeman to a two-level reduction and favorably reduced her Guidelines range. This is so even though (as the government noted at sentencing and again before this Court) Freeman’s ultimate sentence would remain within the recalculated Guideline range if she received nothing more than that reduction. “When a defendant is sentenced under an incorrect Guidelines range—whether or not the defendant’s ultimate sentence falls within the correct range—the error itself can, and most often will, be sufficient to show a reasonable probability of a different outcome absent the error.” Id. at 1345. Because Freeman’s counsel unreasonably failed to argue meritorious objections and advised his client to waive those objections without understanding the gravity of that waiver—and because those objections would have resulted in a reduction of the Guidelines range applicable to Freeman’s sentence—counsel was constitutionally ineffective.

III.

Raising an independent basis for relief, Freeman also argues that her 210-month sentence was substantively unreasonable. Sentencing is generally the province of the district court. See Rita v. United States, 551 U.S. 338, 350–51 (2007). As a result, appellate courts generally defer to a district court’s sentencing decision, and may reverse a sentence only where it is unreasonable, “even if the sentence would not have been the choice of the appellate court.” United States v. Evans, 526 F.3d 155, 160 (4th Cir. 2005); see also Gall *17 v. United States, 552 U.S. 38, 51 (2007). In the Fourth Circuit, “sentences that fall within the Guidelines range are entitled to a presumption of substantive reasonableness.” United States v. Blue, 877 F.3d 513, 519–20 (4th Cir. 2017). Such a presumption, “rather than having independent legal effect, simply recognizes the real-world circumstance that when the judge’s discretionary decision accords with the Commission’s view of the appropriate application of [18 U.S.C.] § 3553(a) in the mine run of cases, it is probable that the sentence is reasonable.” Rita, 551 U.S. at 351. It is accordingly “rebutted by showing that the sentence is unreasonable when measured against the 18 U.S.C. § 3553(a) factors.” United States v. Louthian, 756 F.3d 295, 306 (4th Cir. 2014). We evaluate substantive reasonableness based on “the totality of the circumstances.” United States v. Mendoza-Mendoza, 597 F.3d 212, 216 (4th Cir. 2010). To preserve a claim of substantive reasonableness on appeal, all a defendant must do is, “by advocating for a particular sentence,” communicate to the district court that the sentence is “greater than necessary” in the proceedings below. Holguin-Hernandez v. United States, 140 S. Ct. 762, 767 (2020).

“It has been uniform and constant in the federal judicial tradition for the sentencing judge to consider every convicted person as an individual and every case as a unique study in the human failings that sometimes mitigate, sometimes magnify, the crime and the punishment to ensue.” Koon v. United States, 518 U.S. 81, 113 (1996). These considerations are generally analyzed by district courts within the framework of 18 U.S.C. § 3553(a), which sets out factors district courts must consider in sentencing. They include “the nature and circumstances of the offense and the history and characteristics of the defendant”; “the need for the sentence imposed”; and “the need to avoid unwarranted *18 sentence disparities among defendants with similar records who have been found guilty of similar conduct.” Id. In sentencing Freeman to serve 210 months, the district court did not address sentencing disparities nor fully consider the history and circumstances of the defendant in relation to the extreme length of her sentence.

With regard to sentencing disparities, counsel provides this Court with data obtained from the United States Sentencing Commission’s 2018 Sourcebook of Federal Sentencing Statistics tending to show that Freeman’s sentence is significantly longer than those of similarly-situated defendants. 7 This is so even when considering Freeman’s extremely high (and erroneous) drug weight. Among other disparities, the data show that, in fiscal years 2016-2018, the median sentence for defendants convicted of opioid-related crimes who trafficked ten to fifteen million grams of marijuana-equivalent drug weight was 75 months; the average sentence was about 95 months. That summary includes, according to counsel, “all criminal history levels, including career criminals; all offense levels, including those with offense levels greater than Freeman; and even those with weapons enhancements.” Freeman’s 210-month sentence, counsel notes, is “more than two-and-ahalf times the median sentence for opioid offenders,” and about double the average. Counsel also presents data tending to show that Freeman’s sentence is likewise disparate

7 This Court takes judicial notice of the Sourcebook of Federal Sentencing Statistics data, which is a publication of the United States Sentencing Commission. It is available in print at https://www.ussc.gov/sites/default/files/pdf/research-and-publications/annual- reports-and-sourcebooks/2018/2018-Annual-Report-and-Sourcebook.pdf and in interactive form at https://ida.ussc.gov/analytics/saw.dll?Dashboard.

within both the Fourth Circuit and the District of South Carolina. 8 In its brief, the government offers no substantive response to these disparities raised by counsel, but instead argues only that the Guidelines calculation was correct.

In addition to the disparities raised by Freeman’s counsel, consideration of the § 3553(a) factors also tends to rebut the presumption that Freeman’s sentence is substantively reasonable. Most notably, Freeman was severely addicted to hydrocodone at the time she committed the offenses. The record indicates that she acquired the drugs and sold them principally in order to sustain her own addiction. Freeman and her attorney explained her addiction to the district court, and asked the court for mercy on this basis. But the district court failed to seriously consider Freeman’s addiction as mitigating. In the absence of countervailing evidence such as acts of violence, none of which appears in the record before this Court, Freeman’s severe opioid addiction and her disparate sentence merited a downward variance in this case.

Like oxycodone, hydrocodone is highly addictive. It “is associated with severe psychological or physical dependence” and creates “euphoric effects.” See Schedules of Controlled Substances: Rescheduling of Hydrocodone Combination Products from Schedule III to Schedule II, 79 Fed. Reg. 49,661, 49,665–66, 49,675 (Aug. 22, 2014) (codified at 21 C.F.R. § 1308.13(e)(1)). Like other kinds of opioids, hydrocodone requires users to take progressively more pills in order to feel its effects and avoid painful

8 While defense counsel did not raise the issue of sentencing disparities below, the government on appeal failed to argue that this issue was waived nor has it asked this Court to dispose of the substantive reasonableness argument on that ground. The government has accordingly “waived the waiver,” United States v. Weaver, 66 F. App’x 453, 454 n.1 (4th Cir. 2003), and this Court need not address any waiver before proceeding to the merits of Freeman’s argument.

withdrawal symptoms. The addiction can quickly take over a person’s life. Take, for example, what one attorney wrote in the Texas Bar Journal:

“Lortab filled the void in my life. After taking one a day for the first month or so, I moved up to two a day, three a day, and, before I knew it, I had gone through the three refills remaining on the prescription. When I ran out of refills, I started going to doctors I knew, making up symptoms such as pain or a severe cough so that they would prescribe something containing hydrocodone, the active ingredient in Lortab. I spent all day at work thinking about how I could get my hands on more Lortab. Eventually, my habit got to 30 tablets a day. This pattern continued for the next two years, with my habit eventually reaching 50 tablets a day, taking 10 at a time every six hours or so. If I ever ran out, I would go into horrible withdrawals, with diarrhea, my legs shaking uncontrollably, my nose running, and being unable to sleep or think straight until I either got more drugs or a week or two had passed.”

“Too Young Not to Quit!” 64 Tex. Bar J. 176 (Feb. 2001). Another lawyer wrote of his addiction: “It wasn’t long until I had a full-time dealer selling me 30 Lortabs every two weeks. That number continued to climb. Within five years, I was taking 10 or 12 at a time just to function.” “Living in Darkness,” 76 Ala. Law. 124 (March 2015).

Based on the disparity between her sentence and those of similar defendants, and on the overwhelming record evidence of Freeman’s addiction to opioids, we conclude that Freeman has rebutted the presumption of reasonableness and established that her sentence is substantively unreasonable. To the extent that the court referenced the danger of opioids in sentencing Freeman, it was only to condemn Freeman for selling them. While this was certainly not an improper factor for the district court to consider, it also does not reflect the full picture. And although the district court stated that Freeman was “no doubt a major supplier” of hydrocodone, it failed to consider that the amount that Freeman sold was frequently no more than half of what she was taking herself.

Freeman was first prescribed hydrocodone for an injury as a teenager in 2000, around the time of her graduation from high school. As noted above, there is no evidence in the record that she was ever involved in violent or gang activity, ever used or possessed a firearm, or ever engaged in any other criminal conduct besides conduct related to illegally filling prescriptions for opioids in service of her addiction. And while forging prescriptions for opioids and selling them illegally is reprehensible conduct, Freeman’s offense of conviction also had no identifiable victims. 9

For purposes of this appeal, we need not decide what sentence would be reasonable for Freeman; that is a decision for the district court to make in the first instance. We hold only that a more than 17-year sentence is substantively unreasonable under the circumstances of this case. This unreasonableness is an independent basis for vacating Freeman’s sentence.

• • * *

Accordingly, because her counsel was ineffective and her sentence is substantively unreasonable, we vacate Freeman’s sentence and remand for resentencing consistent with this opinion.

VACATED AND REMANDED

9 It is worth noting that other parties appear to bear significantly greater responsibility for the opioid crisis. See, e.g., Department of Justice, “Justice Department Announces Global Resolution of Criminal and Civil Investigations with Opioid Manufacturer Purdue Pharma and Civil Settlement with Members of the Sackler Family,” (Oct. 21, 2020), https://www.justice.gov/opa/pr/justice-department-announces-global- resolution-criminal-and-civil-investigations-opioid.

QUATTLEBAUM, Circuit Judge, dissenting:

This sad case illustrates the opioid epidemic ravaging our country. Precias Freeman is a victim of this epidemic. As a teenager, she succumbed to the highly addictive nature of opioids in a way that continues to wreak havoc on her life. As a fellow citizen, I am heartbroken over the toll her addiction has levied. But Freeman chose to be a culprit too. By her own admission, she prolifically forged prescriptions to obtain opioids for years— not just for herself, but to sell to others. Whatever role her addiction played, that conduct was plainly criminal and certainly not bereft of “victims.” Maj. Op. at 21. Thus, today, we consider the sentence she received after pleading guilty of possession with intent to distribute two opioids, Hydrocodone and Oxycodone. The majority vacates Freeman’s sentence for two reasons. It concludes that the sentence was substantively unreasonable and that Freeman received ineffective assistance of counsel. Both holdings are unprecedented in our circuit.

First, as to substantive unreasonableness, our decision is indeed remarkable. This is the first time our circuit has held a within-Guidelines sentence substantively unreasonable. And Freeman’s sentence was not just within the range recommended by the Sentencing Guidelines, but at the low endpoint. As such, it was by law presumptively reasonable. Yet, the majority decides that presumption has been rebutted because the district court “failed to seriously consider Freeman’s addiction as mitigating,” Maj. Op. at 19, and because of “the need to avoid unwarranted sentence disparities.” Maj. Op. at 17–18. Respectfully, both reasons are flawed. The district court in fact considered Freeman’s addiction, particularly with respect to Freeman’s motion for admission into the District of South Carolina’s drug *23 court program. The fact that it chose to place more weight on the “devastating effects” from Freeman’s illicit opioid distribution does not come close to abusing the discretion district courts are afforded in sentencing. J.A. 139. We do not have the authority, as an appellate court, to decide the best weight to give each sentencing factor. That is what district courts do. The district court properly considered all the applicable sentencing factors, including Freeman’s history of addiction, and weighed them well within its discretion.

As to an unwarranted sentence disparity, there was none in this case as Freeman was the only defendant, and a nationwide comparison does not reveal any substantial unwarranted disparities. The majority’s attempt at analyzing nationwide sentencing statistics fails to compare apples to apples, exemplifying the dangers of courts conducting independent statistical analyses. Worse, it places an unnecessary and ill-advised burden on district courts to do the same.

Second, regarding ineffective assistance of counsel, my disagreement is primarily procedural. What the majority outlines is compelling. But this is a direct appeal. Thus, we have not even heard Freeman’s counsel’s side of the story. For good reason, courts almost never reach, much less find, ineffective assistance of counsel on direct appeal. Indeed, until today, we never have. Instead, collateral review is the most appropriate procedure for addressing this issue because it allows a court to hear from the counsel whose conduct is alleged to be ineffective. I would not deviate from that sound procedure here.

For these reasons, I dissent on both grounds. I would affirm Freeman’s sentence.

I.

For the first time in our circuit’s history, the majority finds a sentence within the properly calculated Guidelines range substantively unreasonable. A survey of our sister circuits shows that other courts have only rarely done so, and always in response to extraordinarily unusual circumstances. See United States v. Jenkins, 854 F.3d 181, 189 (2d. Cir. 2017) (holding that a 225-month imprisonment followed by 25 years of supervised release was a substantively unreasonable sentence for a possessor of child pornography because the sentence treated the defendant “like an offender who seduced and photographed a child and distributed the photographs and worse than one who raped a child”); United States v. Plate, 839 F.3d 950, 957 (11th Cir. 2016) (holding it substantively unreasonable to condition a prison term solely on a defendant’s ability to pay restitution); United States v. Dorvee, 616 F.3d 174, 183 (2d. Cir. 2010) (reasoning that the district court’s “apparent assumption that [the defendant] was likely to actually sexually assault a child,” which was “unsupported by the record,” but nonetheless “motivated” the sentence, was substantively unreasonable); United States v. Amezcua-Vasquez, 567 F.3d 1050, 1055 (9th Cir. 2009) (holding a fifty-six-month sentence for illegal reentry substantively unreasonable because a sixteen-level enhancement applied to “a decades-old prior conviction” without any adjustment for the magnitude of its staleness). This scarcity should come as no surprise, given the deferential standard of review appellate courts must employ in reviewing sentences. What is surprising, however, is that the majority chose this case to be our circuit’s first.

To get there, the majority strays from the proper standard of review. It then undertakes a quantitative analysis of nationwide sentencing data. But unfortunately, in doing so, several problems with its analysis lead to dubious results. Finally, but perhaps most concerning, our holding today seems to require district courts to conduct their own quantitative inquiries to supplant the data-driven studies that underlie the Sentencing Guidelines.

A.

“Substantive reasonableness examines the totality of the circumstances to see whether the sentencing court abused its discretion in concluding that the sentence it chose satisfied the standards set forth in § 3553(a).” United States v. Hargrove, 701 F.3d 156, 160–61 (4th Cir. 2012) (quoting United States v. Mendoza-Mendoza, 597 F.3d 212, 216 (4th Cir. 2010)). A district court must impose a sentence that is “sufficient . . . to comply with the purposes” of sentencing—retribution, deterrence, incapacitation and rehabilitation. 18 U.S.C. § 3553(a); see Rita v. United States, 551 U.S. 338, 347–48 (2007). “The sentencing court must also consider (1) the nature of the offense and the defendant’s history and characteristics; (2) the kinds of sentences legally available; (3) the advisory sentencing range provided by the Sentencing Guidelines; (4) any relevant policy statement issued by the Sentencing Commission; (5) the need to avoid unwarranted sentence disparities; and (6) the need for restitution.” United States v. Zuk, 874 F.3d 398, 409 (4th Cir. 2017). The district court retains, however, “extremely broad discretion when determining the weight to be given each of the § 3553(a) factors.” United States v. Jeffery, 631 F.3d 669, 679 (4th Cir. 2011). And it need not mechanically tick through each of the *26 sentencing factors so long as they are properly considered. United States v. Johnson, 445 F.3d 339, 345 (4th Cir. 2006). Importantly, when a sentence is within or below the Guidelines range, it carries “a presumption of reasonableness.” United States v. Susi, 674 F.3d 278, 289 (4th Cir. 2012).

The majority’s acknowledgement of this deferential standard of review is incomplete. Although it recites that a sentence must be unreasonable for us to vacate it, the majority omits the “extremely broad discretion” language our precedent affords a district court in determining the weight given to each sentencing factor, instead describing sentencing as “generally the province of the district court.” Compare Jeffrey, 631 F.3d at 679, with Maj. Op. at 16. To the contrary, sentencing is always the province of the district court. That is because it is best positioned to weigh all the sentencing factors in an individual case. Gall v. United States, 552 U.S. 38, 51–52 (2007). As an appellate court, we only serve as a safeguard to the wide discretion afforded the district court.

What’s more, after not fully articulating our standard of review, the majority then fails to adhere to it. It concludes that Freeman’s sentence should be vacated because “the district court did not address sentencing disparities nor fully consider the history and circumstances of [Freeman] in relation to the extreme length of her sentence.” Maj. Op. at 18. Nevertheless, the majority does not find any procedural error with the district court’s sentencing hearing. Rightfully so—the district court addressed Freeman’s arguments and thoroughly articulated its reasons for the sentence, considering all the § 3553(a) factors. Freeman did not raise any sentencing disparities, and her counsel’s presentation focused entirely on the mitigating factors of her addiction, which the district court acknowledged, *27 noting her prior intensive treatment for her opioid addiction. Rather, the majority’s expressed concern that Freeman’s addiction was not fully considered seems to be another way of saying that the district court did not place enough weight on those factors. But that, in effect, is a de novo review. And, of course, “it is not for the Court of Appeals to decide de novo whether . . . the sentence is reasonable.” Gall, 552 U.S. at 59. Instead, we must give “due deference” to the district court’s balancing of the § 3553(a) factors. Id. Freeman’s history of addiction could be a compelling mitigating factor. And it may be true that her sentence is greater than some who have trafficked similar amounts of drugs. However, “[t]he uniqueness of the individual case . . . does not change the deferential abuse-of-discretion standard of review that applies to all sentencing decisions.” Id. at 52. Nor can it here.

B.

In addition to disregarding the proper standard of review, I disagree with the majority’s reasoning on both points it raises with respect to the sentence’s reasonableness. The district court sufficiently considered Freeman’s addiction as mitigating, even if perhaps not to the same extent as the majority would. And I see nothing in the record supporting the majority’s conclusion that there is a sentencing disparity, let alone an unwarranted one.

1.

The majority determines that Freeman’s sentence is substantively unreasonable in part because the district court “failed to seriously consider Freeman’s addiction as mitigating.” Maj. Op. at 19. It scolds the district court for only referencing opioids to *28 “condemn Freeman for selling them” rather than also considering that Freeman consumed a substantial amount of her fraudulently obtained opioids herself. Maj. Op. at 20.

But the record shows the district court did, in fact, consider Freeman’s addiction as mitigating. As § 3553(a)(1) requires, the district court “considered the history and characteristics of [Freeman],” including the fact that she had “a prior history of drug use including marijuana and Lortab.” 1 J.A. 136–37. It noted that Freeman has never “had any real substance abuse treatment, but . . . as a part of her Pretrial Services she was placed in intensive outpatient program.” J.A. 137–38.

In addition to these explicit statements, the context of the hearing is important. The vast majority of the hearing focused on Freeman’s addiction. Freeman’s counsel devoted his entire presentation to Freeman’s debilitating addiction in arguing for her admission to the drug court program, or, in the alternative, a downward variance. Freeman herself gave a moving statement describing how her addiction has controlled her life, including some of her criminal conduct. Even the government admitted it had “a lot of sympathy for the issues that [Freeman’s counsel] has brought out” and that he gave “a fine presentation” illustrating Freeman’s addiction. J.A. 122. The district court’s own statements indicate that it considered these arguments, which focused on Freeman’s addiction, when denying her motion to enter the drug court program or to vary downward. The district court clarified twice with Freeman that she consumed dozens of pills per day. After noting that the drug court program was designed for “defendants who have minimal or no criminal record who

1 Lortab is the brand name for a drug combining hydrocodone and acetaminophen.

have an addiction,” it concluded that Freeman was not fit for the program based upon an “individualized assessment” of the sentencing factors, pointing specifically to Freeman’s criminal history which reflected a “lack of respect” for the law and the serious magnitude of her opioid distribution. J.A. 139–40. Put simply, it is off base to suggest that the district court did not fully consider Freeman’s addiction as mitigating. That was the hearing’s headline.

Making matters worse, the majority’s holding seems to mandate that district courts treat a defendant’s drug addiction as a mitigating circumstance to drug trafficking offenses. As a general rule, I do not question that a district court—in its discretion—may consider drug addiction as a mitigating factor, as the district court did here. A defendant’s addiction can be part of the defendant’s “history and characteristics” under 18 U.S.C. § 3553(a)(1). But no other circuit has mandated that a defendant’s addiction be considered mitigating. 2 Instead, most have determined that addiction may, but not must, be considered a mitigating factor. See United States v. Douglas, 713 F.3d 694, 703 (2d. Cir. 2013) (explaining that sentencing courts may give “compassion” and “due consideration of the relative costs and effectiveness of treatment versus long prison sentences” but need not “turn a blind eye to behavior” showing a defendant is “a poor candidate for treatment or for leniency”); United States v. Garcia, 497 F.3d 964, 972 (9th Cir. 2007) (“[D]istrict courts are not prohibited in all circumstances from considering a defendant’s drug addiction in choosing a reasonable

2 To the extent the majority reaches this conclusion because of the unique nature of the opioid epidemic and the “significantly greater responsibility” that falls on other actors, Maj. Op. at 21 n.9, I would note that “this case is not a vehicle for deciding questions of comprehensive drug policy.” Douglas, 713 F.3d at 703.

sentence.”). Some have even expressed skepticism that addiction is a legitimate reason for varying downward “absent extraordinary circumstances.” United States v. Hodge, 469 F.3d 749, 757 (8th Cir. 2006). And this general principle has been applied equally to defendants addicted to opioids. United States v. Robinson, 892 F.3d 209, 213–16 (6th Cir. 2018). In my view, we should not stray from the approaches of our sister circuits.

Although the majority, if it were sitting as a district court, may have placed greater mitigating weight on Freeman’s addiction, it was not unreasonable for the district court to place less weight on it compared to the harmful public effects from Freeman’s distribution of the same drugs that have taken her life hostage. The district court could have given it little weight. Or it could have given it predominant weight. But that is the district court’s decision—not ours. And absent “a different choice” from Congress, district courts “must struggle with the difficult task of sentencing” opioid users who distribute those same drugs into the community and perpetuate the cycle of addiction. Douglas, 713 F.3d at 703.

2.

The majority also bases its conclusion that Freeman’s sentence is substantively unreasonable on an alleged unwarranted sentencing disparity. To be sure, one factor a sentencing court must consider is “the need to avoid unwarranted sentence disparities among defendants with similar records who have been found guilty of similar conduct.” 18 U.S.C. § 3553(a)(6). Although courts “need not explain every disparity during sentencing,” a “substantial” disparity could be grounds for finding a sentence substantively unreasonable if it is unwarranted. Zuk, 874 F.3d at 412 (holding that sentencing a defendant to time served of 26 months imprisonment when the Guidelines provided for 240 months *31 imprisonment and similar offenders under the same Guidelines and criminal history category were sentenced to an average of 309 months was a substantial disparity and “one of the rare cases” where a sentence was substantively unreasonable). Importantly, however, a sentence that “is more severe than average . . . does not mean that it was unwarranted.” United States v. Rivera-Santana, 668 F.3d 95, 106 (4th Cir. 2012).

a.

At the outset, there was no substantial disparity in Freeman’s sentence. To that point, not a single person involved in this case noticed any sentencing disparity until we directed argument on the issue. Neither Freeman’s counsel nor the government raised any concern at sentencing. And Freeman’s newly appointed counsel on appeal initially submitted an Anders brief arguing there was not any non-frivolous issue to appeal. While we, of course, have the authority to direct argument on new issues—as we did here on the disparity between Freeman’s sentence and other opioid defendants—we should do so with recognition of our distance from district court proceedings. Far from that, today, as set forth below, we contort both the law and basic statistical reasoning to conclude that Freeman’s sentence created an unwarranted disparity that renders her sentence unreasonable.

First, the sentencing data the majority cites do not compare the right variables. Freeman had an offense level of thirty-six with a criminal history level of II, which led to the relevant Guidelines range of 210–240 months. This offense level was calculated based upon the amount of opioids Freeman sold, her obstruction of justice and her failure to accept responsibility. The majority should have compared Freeman’s sentence to those of defendants with the same or similar offense levels and criminal histories. But it did not. *32 Instead, the majority compares Freeman’s sentence only with defendants who trafficked a similar amount of opioids. Ironically, by prioritizing drug weight to the exclusion of offense level and criminal history, the majority’s comparison, no doubt unwittingly, fails to consider the other § 3553(a) factors, which must be considered when the sentencing court treats “every convicted person as an individual and every case as a unique study in the human failings that sometimes mitigate, sometimes magnify, the crime and the punishment to ensue.” Koon v. United States, 518 U.S. 81, 113 (1996). By failing to consider the other factors—obstruction and acceptance of responsibility—the majority compared Freeman with dissimilar offenders. In other words, the majority does not, as required, compare Freeman to “defendants with similar records who have been found guilty of similar conduct.” 18 U.S.C. § 3553(a)(6).

Not surprisingly, some opioid sellers have no criminal history at all. Others have violent pasts. Some sell solely to profit off others’ addictions. Others, like Freeman, may be selling to support their own addiction. Some may fully accept responsibility, assist prosecutors and fully comply with court orders on bond. Others, like Freeman, may abscond while on bond for several months without any contact. 3 These variables greatly impact one’s sentence. But the majority’s analysis does not consider them at all. Put simply, the majority doesn’t compare apples to apples. It compares an apple to all fruits, and decries

3 The majority ably outlines the facts in this case. It is worth noting, however, that Freeman has a substantial history of fleeing authorities. In her own words, Freeman admits that she “just ran” during her six-month absconsion on bond, and the government could not locate her until she was subsequently arrested in a different state. J.A. 132.

that, because the apple looks different than oranges or bananas, something must be wrong with it.

Other courts have had no trouble discrediting such sentencing comparisons. See United States v. Zukerman, 897 F.3d 423, 430 (2d. Cir. 2018) (dismissing “arguments based on aggregated sentencing data” that lump together dissimilar defendants and fail to consider only defendants “with similar records who have been found guilty of similar conduct”) (quoting 18 U.S.C. § 3553(a)(6)); United States v. Willingham, 497 F.3d 541, 544 (5th Cir. 2007) (“[A]verages 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.”). Unfortunately, today we embrace them.

Second, the lone factor the majority considered—the amount of opioids sold—is hardly even predictive of a sentence, making it a poor factor upon which to base a comparison. The graphs provided by Freeman’s counsel barely show a sloping trend-line, and the data points are scattered about with little relation, indicating a weak correlation between drug weight and prison sentence. See Appellant’s Supplemental Br. at 23–24. This weak relationship is unsurprising because, by law, drug weight is just one of many considerations that goes to offense level, which is then paired with a defendant’s criminal history and considered alongside the defendant’s other individualized factors. See 18 U.S.C. § 3553(a). Thus, the graphs on which the majority relies belie its reasoning. 4

For convenience, the graphs are provided here. They are also attached as an 4

addendum in full-size for reference.

Figure 1: Prison sentence (in months) versus marijuana equivalent drug weight (in grams) of opioids sold, nationwide, FY2016–18.

Figure 2: Prison sentence versus marijuana equivalent drug weight (in grams) of opioids sold, nationwide, criminal history I-III, offense level 30-60, no weapons enhancement, FY2016–18.

Third, even using drug weight as the only variable—which again presents an incorrect comparison—the graphs supplied by Freeman’s counsel confirm that her sentence is not unusually high. Recall that the district court sentenced Freeman to 210 months imprisonment for trafficking the marijuana-equivalent of just over ten million grams. Although that is greater than the median sentence of 75 months and average sentence of 95 months the majority identified using its criteria, the graphs nonetheless show that at least five defendants who trafficked less opioids received sentences of at least 200 months, and three received sentences greater than Freeman’s. So even if Freeman’s sentence is at the high end when solely considering the amount of opioids she sold, it is by no means significantly disparate. Thus, even using the majority’s metrics, Freeman’s sentence is not a lonely outlier—it has company.

Fourth, a careful look behind the statistics upon which the majority relies is quite revealing. The majority reasons that Freeman’s sentence was disparate because “the data show that, in fiscal years 2016-2018, the median sentence for defendants convicted of opioid-related crimes who trafficked ten to fifteen million grams of marijuana-equivalent drug weight was 75 months; the average sentence was about 95 months.” Maj. Op. at 18. Those statistics are based upon eighteen sentences. Fifteen of the sentences, importantly, benefited from an acceptance of responsibility credit. Not surprisingly, since Freemen did not receive such a credit, her sentence is higher than those sentences. The other three sentences, like Freeman’s, did not benefit from an acceptance of responsibility credit. Those three defendants were sentenced to 188, 216, and 262 months in prison. When compared to those three sentences, Freeman’s sentence is not at all unusually high—it is *36 smack dab in the middle. Thus, a look into the details of the data confirms that Freeman’s sentence was not disparate.

All of this illustrates the dangers of statistics. Without a standardized policy and tools necessary to consider quantitative data, mistakes are easily made. Numbers that on their surface might appear to suggest one thing may actually suggest the opposite. The judiciary, as currently constructed, is not institutionally capable of deploying statistical analysis. After all, that is not our job. We apply the law to the facts.

Fortunately, other institutions are quite capable of these types of analyses. “[T]he Commission fills an important institutional role: It has the capacity courts lack to ‘base its determinations on empirical data and national experience, guided by a professional staff with appropriate expertise.’” Kimbrough v. United States, 552 U.S. 85, 108–09 (2007) (quoting United States v. Pruitt, 502 F.3d 1154, 1171 (10th Cir. 2007) (McConnell, J., concurring)). That is why the Sentencing Guidelines should be our barometer for promoting nationwide sentencing uniformity. That is not to say sentencing courts cannot, or even should not, ever consult national sentencing statistics. But if they do, they should do so with extreme care.

Contrary to the majority’s conclusions, there is no substantial disparity in Freeman’s sentence compared to similarly-situated defendants. Zuk, 874 F.3d at 412. A sentence within the Guidelines range is presumed reasonable. Freeman’s sentence was at the lowest end of the Guidelines range. Thus, Freeman’s sentence is not only presumed reasonable, it is also hardly likely to create a significant disparity because it is at the lowest point in the Guidelines range. Any alleged disparity the majority points to pales in comparison to those *37 other circuits have found troubling. See United States v. Stock, 685 F.3d 621, 629–30 (6th Cir. 2012) (determining that a sentence was “an outlier when compared to other sentences . . . even when that comparison is limited to other criminal-history-category-VI offenders” because it “was nearly twice the top of the (erroneously-high) Guidelines range” and “by far the longest sentence imposed” for that offense in the last few years and “possibly ever”).

b.

Even if there were compelling evidence of a sentencing disparity, which there is not, that alone does not make a sentence substantively unreasonable. It must be an “unwarranted sentence disparity.” 18 U.S.C. § 3553(a)(6) (emphasis added). The fact that Freeman’s sentence is higher than average when compared to other defendants who trafficked a similar amount of opioids does not make the disparity unwarranted. “[T]hese statistics alone do not show that [a] sentence was ‘unwarranted.’” United States v. Snyder, 865 F.3d 490, 502 (7th Cir. 2017). To the contrary, it is entirely warranted because that comparison does not account for Freeman’s criminal history, obstruction of justice or failure to accept responsibility—all factors that are baked into the calculation of the applicable Guidelines range. 5 The Guidelines, for that reason, are a far better benchmark. “[A] reviewing court’s concern about unwarranted disparities is at a minimum when a sentence is within the Guidelines range.” United States v. Willingham, 497 F.3d 541, 545 (5th Cir. 2007). The Supreme Court has explained why: the “avoidance of unwarranted

5 To be clear, I do not fault Freeman’s counsel for making these comparisons, as we instructed her to do so after she already determined there were not any non-frivolous issues meriting appeal in her opening Anders brief. In fact, I commend both counsel in the case for the manner in which they have presented the issues we consider today.

disparities was clearly considered by the Sentencing Commission when setting the Guidelines ranges.” Gall, 552 U.S. at 54. “The federal Sentencing Guidelines were developed based on an extensive statistical analysis of sentencing patterns and the variables affecting them.” 6 See Adi Leibovitch, Punishing on a Curve, 111 Nw. U. L. Rev. 1205, 1272 (2017). Therefore, if the district court “correctly calculated and carefully reviewed the Guidelines range, [it] necessarily gave significant weight and consideration to the need to avoid unwarranted disparities.” Gall, 552 U.S. at 54. The district court below did just that. Id.; see also United States v. Boscarino, 437 F.3d 634, 638 (7th Cir. 2006) (“A sentence within a properly ascertained range . . . cannot be treated as unreasonable by reference to § 3553(a)(6).”), cert. denied, 551 U.S. 1161 (2007).

In the past, we have been “unwilling to isolate a possible sentencing disparity to the exclusion of all the other § 3553(a) factors.” Rivera-Santana, 668 F.3d at 106 (internal quotation marks omitted). Yet that is what we do today. In my view, we should not. The district court’s focus on the other sentencing factors—the need to deter, protect the public and promote respect for the law—was not misplaced and certainly not an abuse of discretion.

3.

6 I recognize that the Commission departed from its traditional empirical approach when setting the Guidelines for drug-trafficking offenses. Instead, U.S.S.G. § 2D1.1(c) “use[s] a drug quantity table based on drug type and weight to set base offense levels for drug-trafficking offenses.” Kimbrough, 552 U.S. at 96. These determinations were initially influenced in large part by the Anti-Drug Abuse Act of 1986, 100 Stat. 3207. This difference, however, does not change the fact that these Guidelines reflect a legislative judgment and “the initial benchmark” of a sentence in furtherance of the goal of national sentence uniformity. Kimbrough, 552 U.S. at 108 (internal citation omitted).

There is a final problem with the majority’s decision. Stepping back from this case, our opinion forces an unnecessary burden upon district courts. Our holding seems to suggest district courts must conduct their own statistical analyses, sua sponte, when determining a sentence. This exponentially compounds the institutional problem of courts applying quantitative analyses they are ill-equipped to perform. And it also imposes a significant, and unnecessary, burden on district courts. While I would not hesitate to support requirements that increase the workload of district courts if doing so was required by law or even represented a positive advancement, requiring the type of review our decision seems to suggest is neither.

Even if we insist district courts plow forward in statistical analyses, how should they do so? What statistics exactly must the district court consider? How narrowly defined must the comparison group be? What is considered too disparate of a sentence? And how should we even review such an analysis? The majority gives no indications how to proceed. 7

7 There may be good policy reasons for district courts to consider the relevant sentencing distributions of similarly-situated defendants in their own and other jurisdictions. See, e.g., Adi Leibovitch, Punishing on a Curve, 111 Nw. U. L. Rev. 1205 (2017) (arguing that judges sentence a given offense more harshly when their caseloads contain relatively milder offenses and vice-versa). I make no comment on this question. But doing so requires “policymakers to make a straightforward choice about the relevant universe for comparison.” Id. at 1252 (emphasis added). And if the decision is made to do so, there are a myriad of ways that Congress or the Commission could choose to empower courts to consider sentencing distributions. For example, the Commission could provide a “curve according to which sentences should be standardized.” Id. at 1254. Or perhaps Congress could decide to equip appellate courts to employ meaningful data-driven sentence reviews to ensure greater uniformity. See Joshua M. Divine, Booker Disparity and Data- Driven Sentencing, 69 HASTINGS L.J. 771 (2018). The point is—that is not our choice. It is Congress’s. Someone needs to set the rules, and that is inherently a legislative, not a judicial, function.

For good reason, other courts have explicitly rejected this type of requirement. The Second Circuit persuasively addressed this issue in United States v. Irving, 554 F.3d 64, 75–76 (2d Cir. 2009). There, the defendant argued the district court should have considered local and national sentencing statistics for similarly situated offenders. Id. at 75. In no uncertain terms, the Second Circuit rejected this argument. Id. at 76 (“The district court was not required to consult these statistics.”). A contrary holding, it reasoned, would require district courts to consult statistics that “provide[] no assurance of comparability because [they] do not distinguish between defendants” within a given offense. Id. A concern over unwarranted disparities, moreover, is already given “significant weight and consideration” when the district court properly considers the Guidelines. Id. (quoting Gall, 552 U.S. at 599). Our new contrary approach—unlike Freeman’s sentence—is an unfortunate outlier.

II.

The majority also breaks new ground today by reversing a sentence on direct appeal for ineffective assistance of counsel. I share many of the same concerns the majority ably outlines. But I cannot conclude the record “conclusively” demonstrates that “counsel did not provide effective representation.” United States v. Benton, 523 F.3d 424, 435 (4th Cir. 2008) (quoting United States v. Richardson, 195 F.3d 192, 198 (4th Cir. 1999)). Until today, we have never found ineffective assistance of counsel on direct appeal. See United States v. Brown, 757 F.3d 183, 191 (4th Cir. 2014) (“[W]e routinely decline to address on direct appeal a criminal defendant’s contention that counsel has performed in an ineffective *41 manner . . . .”). This record should not compel us to make this case the first. I would wait to evaluate this claim on collateral review with the benefit of a fully developed record.

A.

Whether counsel provided ineffective assistance is a mixed question of law and fact which we review de novo. Smith v. Moore, 137 F.3d 808, 817 (4th Cir. 1998). Typically, however, a defendant raises an ineffective assistance of counsel claim collaterally by a motion under 28 U.S.C. § 2255. United States v. King, 119 F.3d 290, 295 (4th Cir. 1997). Doing so allows for adequate development of the record. Id. An ineffective assistance of counsel claim is only cognizable on direct appeal, like here, if the record conclusively demonstrates that the lawyer failed to provide effective representation. See Benton, 523 F.3d at 435. It must be apparent “in the trial record itself that the defendant was not provided . . . effective representation.” United States v. Mandello, 426 F.2d 1021, 1023 (4th Cir. 1970).

To prove counsel was constitutionally ineffective, a defendant must show that (1) counsel’s performance fell below an objective standard of reasonableness, and (2) the deficient representation prejudiced the defendant. Strickland v. Washington, 466 U.S. 668, 687 (1984). As to the first prong, counsel’s “ignorance of a point of law that is fundamental to his case combined with his failure to perform basic research on that point is a quintessential example of unreasonable performance.” Hinton v. Alabama, 571 U.S. 263, 274 (2014) (per curiam). At the same time, however, the Supreme Court has directed lower courts to “indulge a strong presumption that counsel’s conduct falls within the wide range of reasonable professional assistance” when evaluating counsel’s performance under the *42 first prong. Strickland, 466 U.S. at 694. Thus, “judicial scrutiny of counsel’s performance must be highly deferential,” Roe v. Flores-Ortega, 528 U.S. 470, 477 (2000) (citation and alteration omitted), and from the standpoint of whether counsel’s perspective at the time would indicate the challenged action “might be considered sound . . . strategy.” Strickland, 466 U.S. at 689 (quoting Michel v. Louisiana, 350 U.S. 91, 101 (1955)).

B.

At sentencing, Freeman’s counsel withdrew objections to the presentence report (“PSR”) that may have lowered Freeman’s offense level, and thus the Guidelines range. Prior to the sentencing hearing, Freeman’s counsel had lodged eight objections to the PSR, at least three of which related to either the drug weight Freeman trafficked or her obstruction of justice enhancement. At the hearing, however, Freeman’s counsel withdrew all objections. Troublingly, her counsel indicated to the court that “none of those objections reduce the number that is relevant to this Court.” J.A. 108. As the majority rightly notes, that is not true. At stake were potentially seven offense levels—two related to drug weight, two for obstruction of justice and three for failure to accept responsibility. Success on any of those three objections would have reduced the Guidelines range, and success on all three would have resulted in a Guidelines range of 97–121 months, compared to her conceded range of 210–240 months. Freeman’s counsel was wrong to suggest to the court that the objections were irrelevant to the Guidelines calculation.

Instead of pursuing these objections, Freeman’s counsel decided to focus solely on the motion to enter the drug court program. He made a compelling argument in his motion and presentation at the sentencing hearing that Freeman suffered from a debilitating *43 addiction and that all her criminal conduct was tied to that addiction. In this presentation, counsel focused on Freeman’s positive and sympathetic traits—including the fact that she has several young children, a supportive family and a strong desire to conquer her addiction.

Although the majority rightly notes that counsel’s decision to put all its eggs in the drug court basket may have been a poor strategy, it nonetheless seems quite plausible that it was a tactical decision to withdraw the objections. The withdrawal could more broadly be viewed as a strategic choice to avoid focusing on Freeman’s criminal activity at the sentencing hearing, including her prolific fraudulent prescription filling and months-long absence from the jurisdiction incommunicado, and to instead focus on her mitigating traits. This strategy is quintessential in sentencing hearings. The government was prepared to call several witnesses if Freeman chose to litigate the objections, and it may not have been unreasonable for Freeman’s counsel to consider that this testimony may irreparably alter the sympathetic narrative he wished to convey to the district court. See J.A. 109–10 (“I have got Ms. Rogers here from Probation as you mentioned; I have a DHEC inspector, Mr. Thomason, who interviewed her along with Rachel Richmond back in October 2016; and I have Adam Roberson with DEA Diversion, the Federal case agent, who interviewed Ms. Freeman as well prior to her absconding. And to the extent we need testimony, all of these individuals are ready to go.”). Thus, under the deferential standard we must consider counsel’s conduct, I cannot say it is impossible that his strategy was reasonably competent.

In fact, our opinion today problematically holds that such a strategy “would never be an appropriate justification for waiving a meritorious [objection].” Maj. Op. at 14. This *44 cannot be true—in fact, it flatly contradicts precedent. It is true that “counsel may be constitutionally required to object when there is relevant authority strongly suggesting that a sentencing enhancement is not proper.” United States v. Carthorne, 878 F.3d 458, 466 (4th Cir. 2017) (emphasis added). But in general, “counsel may have a strategic reason for not raising a particular objection.” Id. at 467. Consistent with Supreme Court precedent, we previously held that “the failure to raise an objection that would be apparent from a thorough investigation is a significant factor in evaluating counsel’s performance” only “in the absence of such a [strategic] reason.” Id.; see also Wiggins v. Smith, 539 U.S. 510, 526 (2003) (finding ineffective assistance of counsel when counsel’s failure to thoroughly investigate resulted from inattention rather than “reasoned strategic judgment”); Strickland, 466 U.S. at 691 (“In any ineffectiveness case, a particular decision not to investigate must be directly assessed for reasonableness in all the circumstances, applying a heavy measure of deference to counsel’s judgments.”). The majority, however, requires all meritorious objections to be made, even though there may be legitimate strategic reasons to waive some. After all, a meritorious objection is not necessarily likely to succeed, and the costs of pursuing one may outweigh the perceived benefits.

But, of course, we do not know why counsel advised Freeman to waive the objections. It is impossible to discern from the record alone. That is why we need to develop a record that includes testimony from Freeman’s counsel. “[I]t would be unfair to adjudicate the issue without any statement from counsel on the record.” United States v. DeFusco, 949 F.2d 114, 120 (4th Cir. 1991). Without more evidence as to why Freeman’s counsel chose to take certain actions, “it is impossible to make a reasoned judgment as to *45 whether or not representation was ineffectual.” Id. at 120–21 (internal citation and quotation marks omitted). 8

Because we do not know exactly why counsel advised Freeman to withdraw the objections, we should leave this claim for collateral review. It is “best left to collateral review” so Freeman can develop an adequate record, which could shed light on why Freeman’s counsel advised her to withdraw the objections, and how the court may have ruled on the objections if the government’s witnesses were to testify. United States v. Hoyle, 33 F.3d 415, 418 (4th Cir. 1994); see also Massaro v. United States, 538 U.S. 500, 504–06 (2003) (“[I]n most cases a motion brought under § 2255 is preferable to direct appeal for deciding claims of ineffective assistance.”). Having Freeman pursue this claim on a § 2255 motion would allow her to produce this evidence, and it would prevent us from speculating about counsel’s motives and capabilities.

8 Even if we were to assume counsel’s performance fell below the constitutional standard, it is unclear if it prejudiced Freeman. When evaluating the second prong of Strickland, the question is “whether ‘there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.’” Padilla v. Kentucky, 559 U.S. 356, 366 (2010) (internal citation and quotation marks omitted). Had Freeman’s counsel only succeeded in lowering the drug weight, Freeman’s offense-level would have dropped by two points, and 210 months would have still been within the Guidelines, albeit the high point.

The district court stated that it would have still maintained the same sentence as an alternate variant sentence even if the Guidelines range differed. Although this type of statement cannot automatically immunize the sentence from prejudice when counsel’s performance could have resulted in different facts before the court, it must mean something. We have previously held that error may be harmless if it is clear that “the district court would have reached the same [sentence] even if it had” decided a Guidelines issue differently. United States v. Gomez, 690 F.3d 194, 203 (4th Cir. 2012). Here, I would conclude the statement cuts against a clear finding of prejudice.

III.

I have great sympathy for Freeman’s circumstances. Her story reflects failures in our community. One could argue her sentence does not reflect sound policy. But that does not make it unreasonable under the law. And while the record is concerning regarding the effectiveness of counsel Freeman received, it does not conclusively demonstrate a failure to meet the constitutional bar at this juncture. I dissent.

Figure 1: Prison sentence versus marijuana equivalent drug weight of opioids sold, nationwide, FY2016–18.

Figure 2: Prison sentence versus marijuana equivalent drug weight (in grams) of opioids sold, nationwide, criminal history I-III, offense level 30-60, no weapons enhancement, FY2016–18.