In the

United States Court of Appeals For the Seventh Circuit ____________________ Nos. 22-2994 & 23-1461 UNITED STATES OF AMERICA, Plaintiff-Appellee, v.

CHRISTOPHER ALLEN YATES and SHAWN THOMAS CONNELLY, Defendants-Appellants. ____________________

Appeals from the United States District Court for the Central District of Illinois. Nos. 4:20-cr-40017-7 & 4:20-cr-40017-4 — Sara Darrow, Chief Judge. ____________________

ARGUED JANUARY 8, 2024 — DECIDED APRIL 11, 2024 ____________________

Before WOOD, SCUDDER, and ST. EVE, Circuit Judges. ST. EVE, Circuit Judge. Christopher Yates and Shawn Con- nelly appeal their sentences following convictions for conspir- ing to distribute methamphetamine. Both challenge the dis- trict court’s finding that the conspiracy involved at least 737.1 grams of “ice” methamphetamine, meaning methampheta- mine that was at least 80% pure. Yates argues that the govern- ment failed to meet its burden of proving the purity of all that 2 Nos. 22-2994 & 23-1461

methamphetamine, having only tested a small, unrepresenta- tive amount. Connelly asserts that the court should not have relied on his coconspirators’ statements to calculate the total drug weight, and that the full weight was not reasonably fore- seeable to him. We vacate Yates’s sentence and remand. The Guidelines allow district courts to engage in some degree of estimation when determining drug quantity and purity, but the govern- ment must supply reliable evidence making that approxima- tion reasonable. Because we find such evidence lacking here, Yates is entitled to resentencing. We affirm Connelly’s sen- tence. I. Background In December 2020, a grand jury indicted Chistopher Yates and Shawn Connelly, along with six coconspirators, on charges of conspiring to distribute at least 50 grams of actual methamphetamine. See 21 U.S.C. §§ 841(a)(1), 841(b)(1)(A), 841(b)(1)(C), 846. Connelly also faced a charge of distributing a mixture and substance containing a detectable amount of methamphetamine. See 21 U.S.C. §§ 841(a)(1), 841(b)(1)(C); 18 U.S.C. § 2. Yates and Connelly pleaded guilty without a plea agree- ment. The following facts come from the presentence investi- gation reports (“PSRs”) prepared before their sentencings. The conspiracy operated out of Macomb, Illinois, and lasted thirteen months, from January 2019 to February 2020. Yates supplied its methamphetamine. At first, he purchased the drugs from an unknown source in Joliet, Illinois, with al- leged Mexican cartel connections. But law enforcement Nos. 22-2994 & 23-1461 3

arrested that supplier sometime around December 2019, forc- ing Yates to seek out a new source. Yates would bring the drugs to his home in Macomb and front them to the other members of the conspiracy to distrib- ute. Connelly was among the distributors. A. Controlled Buys The government conducted one seizure and nine con- trolled buys from members of the conspiracy during its inves- tigation. Most of this activity took place over one week in Sep- tember 2019, when the government executed six controlled buys and obtained more than 144 grams of methampheta- mine (out of 158 grams of methamphetamine obtained in to- tal). The government tested five of these samples, amounting to 141.3 grams of methamphetamine, for purity. Each test identified d-methamphetamine hydrochloride with a purity of 99% or above. A table summarizing the seized metham- phetamine appears below. Date Seized Quantity (g) Purity September 19, 2019 3.496 99% September 20, 2019 2.927 99% September 21, 2019 17.9 100% September 21, 2019 110.2 100% September 23, 2019 3.3 Untested September 25, 2019 6.8 100% October 7, 2019 3.5 Untested October 7, 2019 4 Untested December 11, 2019 6.1 Untested January 22, 2020 0.5 Untested 4 Nos. 22-2994 & 23-1461

B. Drug Quantities The PSRs used coconspirator admissions to calculate the total drug weight attributable to the conspiracy. They relied on three statements in particular: • Amber Phelps’s admission that she obtained at least 481.95 grams of methamphetamine from Yates. • Phelps’s admission that she was present three or four times when Yates sold one ounce (28.35 grams) of methamphetamine to an unindicted individual, for a total of at least 85.05 grams of methampheta- mine. • Jeanna Rechkemmer’s admission that she received a total of 6 to 12 ounces (170.1 to 340.2 grams) of methamphetamine from Yates over the course of the conspiracy. Although evidence suggested the conspiracy involved an even greater amount of methamphetamine, the PSRs deter- mined based on these statements that the conspiracy was re- sponsible for conspiring to distribute at least 737.1 grams of “actual” methamphetamine. That calculation resulted in both defendants having a base offense level of 34, which dropped to 31 after taking into account their timely acceptance of re- sponsibility. See U.S.S.G. § 2D1.1(c)(3). Yates’s criminal his- tory category (III) led to a Guidelines range of 135 to 168 months’ imprisonment. Connelly’s criminal history category (IV) produced a Guidelines range of 188 to 235 months’ im- prisonment. Nos. 22-2994 & 23-1461 5

C. Sentencing 1. Yates The main dispute at Yates’s sentencing was the classifica- tion of the methamphetamine attributable to him as “ice” or “actual” (i.e., pure) methamphetamine, as opposed to its ge- neric variant, which the Guidelines treat less severely. Yates argued that the government had not met its burden of estab- lishing the purity of all 737.1 grams of methamphetamine in- volved in the conspiracy since it had only tested 141.3 grams. The district court rejected that argument and found the tested samples reliably represented the purity of the conspir- acy’s methamphetamine as a whole. It noted the tested sam- ples came from controlled buys with three different cocon- spirators, occurred on separate occasions, and involved vary- ing amounts—all of which suggested to the court that the con- spiracy was not diluting its methamphetamine. The district court also relied on the consistency in purity levels between the tested samples, and the fact that they rep- resented a sizeable proportion—nearly twenty percent—of the total amount of methamphetamine attributed to the con- spiracy. According to the district court, twenty percent was “an appropriate percentage to have the necessary level of con- fidence to rely upon the tested substances to be … a good rep- resentative sample.” Having found Yates responsible for 737.1 grams of “ice” methamphetamine, the district court adopted the PSR’s Guidelines calculations. It sentenced Yates to 168 months in prison—the top of his Guidelines range. 6 Nos. 22-2994 & 23-1461

2. Connelly Connelly raised two objections to the PSR at his sentencing hearing. He argued that the court could not rely on the state- ments of his coconspirators, Amber Phelps and Jeanna Rech- kemmer, for purposes of determining the amount of metham- phetamine handled by the conspiracy, since they both had ad- mitted to lying in statements to law enforcement. He also con- tended that the full drug weight attributed to the conspiracy was not reasonably foreseeable to him. The district court denied both objections. The court con- cluded that although Phelps and Rechkemmer had admitted to misleading law enforcement, their statements as to drug quantity were reliable because of other corroborating evi- dence in the record. It further found Connelly’s involvement in the conspiracy was such that its entire drug weight was foreseeable to him. As it had with Yates, the district court adopted the PSR’s recommendations in full. It sentenced Connelly to 188 months’ imprisonment. Both defendants appeal their sentences. II. Analysis On appeal, Yates again challenges the district court’s find- ing that the methamphetamine attributable to the conspiracy was “ice” methamphetamine. Connelly once more attacks the district court’s calculation of the total drug weight attributa- ble to him. Our review of both claims of error is the same. The government must prove facts including drug purity and quantity by a preponderance of reliable evidence. United States v. Moore, 52 F.4th 697, 700 (7th Cir. 2022); United States v. Rollerson, 7 F.4th 565, 570 (7th Cir. 2021). We review a Nos. 22-2994 & 23-1461 7

district court’s reliability determinations for an abuse of dis- cretion and its factual findings—including drug quantity cal- culations—for clear error. United States v. Jones, 56 F.4th 455, 506 (7th Cir. 2022); Moore, 52 F.4th at 700. A. Yates The Guidelines distinguish between three different vari- ants of methamphetamine: “methamphetamine,” “metham- phetamine (actual),” and “ice” methamphetamine. See U.S.S.G. § 2D1.1(c). “Methamphetamine”—what we will call the generic variety—refers to “the entire weight of any mix- ture or substance containing a detectable amount” of meth- amphetamine. U.S.S.G. § 2D1.1, Notes to Drug Quantity Table (A). “Actual” methamphetamine refers to the weight of the controlled substance itself contained in the mixture or sub- stance. U.S.S.G. § 2D1.1, Notes to Drug Quantity Table (B). Fi- nally, “ice” methamphetamine “means a mixture or sub- stance containing [methamphetamine] of at least 80% purity.” U.S.S.G. § 2D1.1, Notes to Drug Quantity Table (C). These variants carry meaningful sentencing conse- quences. The Guidelines treat quantities of “actual” and “ice” methamphetamine—i.e., any methamphetamine above 80% purity—as ten times the weight of the generic variant. See United States v. Bostock, 910 F.3d 348, 350 (7th Cir. 2018). So, a defendant responsible for 45 kilograms of generic metham- phetamine receives the same base offense level as a defendant responsible for only 4.5 kilograms of “ice” or “actual” meth- amphetamine. See U.S.S.G. § 2D1.1(c). “In short, purity mat- ters for methamphetamine.” United States v. Carnell, 972 F.3d 932, 939 (7th Cir. 2020). 8 Nos. 22-2994 & 23-1461

It comes as no surprise then that purity came to the fore during Yates’s sentencing. The district court found that all 737.1 grams of methamphetamine for which the conspiracy (and therefore Yates) was accountable was “ice” methamphet- amine. To reach that conclusion, the district court extrapo- lated from the 141.3 grams of pure or nearly pure metham- phetamine the government seized and tested during its inves- tigation. Although the district court acknowledged that these tested samples reflected just a “snapshot” of the conspiracy— they came from five controlled buys conducted only within a one-week period—it nevertheless concluded they painted a reliable and accurate picture of the rest of the conspiracy’s supply. On appeal, Yates contends that any act of extrapolation from these samples constitutes error, and that testing all the methamphetamine attributed to a defendant is the only relia- ble way to determine its purity. Failing that, Yates alterna- tively maintains that the tested samples were not reliably rep- resentative of the whole and cannot support the district court’s extrapolation. We consider each position in turn. 1. Testing We recently evaluated the reliability of certain types of ev- idence in proving methamphetamine purity in Carnell. 972 F.3d 932. The government there had not tested any of the methamphetamine at issue and sought to meet its burden us- ing evidence of how dealers and users described the prod- uct—as “ice,” “high quality,” “crystalline,” “glass-like,” and so on. We found that kind of unscientific say-so of little value, declining to borrow from our toolkit for distinguishing be- tween cocaine base and crack cocaine. Id. at 940. Unlike with those drugs, the Sentencing Guidelines adopt a purity-based Nos. 22-2994 & 23-1461 9

approach for classifying methamphetamine variants, which is “both subject to a rigid definition and testable.” Id. at 941. Re- spect for that approach led us to hold that “the experience of users, dealers and law enforcement officers, without more,” did not satisfy the government’s burden of proving that meth- amphetamine was at least 80% pure. Id. We expressly left “for another day” in Carnell the question whether the government must test all methamphetamine at- tributable to the defendant. Id. at 944. Yates now argues that we should adopt that rule. We decline to do so. The government must prove metham- phetamine purity under the Guidelines only by a preponder- ance of the evidence. Id. at 938 (noting that the government’s burden at sentencing is “substantially lower than at trial,” re- quiring only that it “prove that the methamphetamine at- tributed to the defendant is more likely than not methamphet- amine as described in the Guidelines” (quotation marks omit- ted)). Meeting that standard does not require the government to test every last gram of methamphetamine. Yates’s rule would effectively raise the standard to one of near certainty— far beyond what due process and our caselaw require. See United States v. Watts, 519 U.S. 148, 156 (1997); U.S.S.G. § 6A1.3, cmt. (“The Commission believes that use of a prepon- derance of the evidence standard is appropriate to meet due process requirements and policy concerns in resolving dis- putes regarding the application of the guidelines to the facts of a case.”). Yates’s all-or-nothing testing proposition would also lead to absurd results. It would, for example, preclude a district court from finding a defendant responsible for “actual” meth- amphetamine, even if the defendant himself has stipulated to 10 Nos. 22-2994 & 23-1461

its purity. As even Yates conceded at oral argument, that re- sult is both illogical and untenable. Carnell does not require otherwise. While we rejected that certain kinds of evidence could be sufficient to satisfy the gov- ernment’s burden, we expressly declined to “rul[e] out the possibility that there may be other evidence of purity” besides lab reports. Carnell, 972 F.3d at 943–44 (“This is not to say that a lab report is always needed to meet the burden required.”). For instance, assuming the science backs up the existence of a relationship between the purity and the appearance of meth- amphetamine, “the government could provide evidence con- necting the visual description of the methamphetamine to the purity.” Id. at 944; see also United States v. Williams, 19 F.4th 374, 380 (4th Cir. 2021) (“[C]ertainly, lab results of the drugs from the conspiracy at issue often provide the best evidence that the conspiracy, in fact, involves Ice. But we cannot con- clude that such evidence is required in every case.”). We also did nothing to disparage the validity of testing a representative sample of methamphetamine, which courts have repeatedly endorsed as a sufficiently reliable method of proving drug pu- rity. See, e.g., United States v. Dinh, 920 F.3d 307, 313 (5th Cir. 2019) (“We have also held that sentencing courts are permit- ted to extrapolate the nature and quantity of drugs involved in an offense based on lab reports that tested only a sample of the overall quantity.” (citations omitted)); United States v. Hou- ston, 338 F.3d 876, 879 (8th Cir. 2003) (“The government may prove the total quantity of actual methamphetamine in a se- ries of transactions by testing the purity of a seized quantity and applying the percentage of actual methamphetamine in the tested quantity to the unrecovered quantities.”); see also 2 McCormick on Evid. § 213 (8th ed. 2020) (“Real evidence con- sisting of samples drawn from a larger mass are also generally Nos. 22-2994 & 23-1461 11

held admissible to authenticate the total mass, subject to the foregoing requirements pertaining to real evidence generally and to the further requirement that the sample be established to be accurately representative of the mass.”). We therefore have little trouble concluding that the gov- ernment need not test all the methamphetamine attributable to a defendant to meet its burden of demonstrating purity by a preponderance of the evidence. But that does not fully re- solve Yates’s appeal. It brings us only to the next question: whether the government met its burden here. 2. Proving Purity While the Guidelines “permit some amount of reasoned speculation and reasonable estimation by a sentencing court” in determining drug quantity, the district court’s powers of approximation are not limitless. United States v. Hollins, 498 F.3d 622, 631 (7th Cir. 2007) (emphasis in original) (quotation marks omitted). District courts must base their estimates “on evidence possessing a sufficient indicia of reliability.” United States v. Miller, 834 F.3d 737, 741 (7th Cir. 2016) (quoting United States v. Durham, 211 F.3d 437, 444 (7th Cir. 2000)). That rule applies with equal force to drug purity determi- nations. District courts are free to extrapolate from a series of test samples, so long as the government has shown that those samples reliably support the court’s ultimate conclusion. See United States v. Titus, 78 F.4th 595, 601 (3d Cir. 2023) (holding that extrapolation is a “way[] for the government to satisfy its burden of proof”). The evidence comprising the sample must be reliable, and the sample size must be sufficient—whether in size, temporal scope, or both—to reliably serve as a basis for extrapolation. United States v. Self, 681 F.3d 190, 202 (3d 12 Nos. 22-2994 & 23-1461

Cir. 2012) (“When a defendant ‘challenges a drug quantity es- timate based on an extrapolation from a test sample, the gov- ernment must show, and the court must find, that there is an adequate basis in fact for the extrapolation and that the quan- tity was determined in a manner consistent with accepted standards of reliability.’” (quoting United States v. McCutchen, 992 F.2d 22, 25 (3d Cir. 1993))); see also United States v. Jackson, 470 F.3d 299, 310–11 (6th Cir. 2006); United States v. Scalia, 993 F.2d 984, 989 (1st Cir. 1993)). Put simply, there must be evi- dence that the part reliably represents the whole. A court’s evaluation of the representativeness of a given sample is a re- liability determination we review for abuse of discretion. See Carnell, 972 F.3d at 943. The district court concluded that the five tested samples— the only evidence of purity besides the types of evidence we found insufficient in Carnell—reliably and accurately repre- sented the conspiracy’s total drug weight. We find instead that the government did not meet its burden here. The tested samples represented a small fraction—one week—of the conspiracy’s thirteen-month duration. One week out of fifty-six is not representative without evidence that the conspiracy consistently dealt with methamphetamine of similar purity. See United States v. Rivera-Maldonado, 194 F.3d 224, 231 (1st Cir. 1999) (“Generally speaking, the smaller the sampling, the less reliable the resulting probability esti- mate.” (citing David H. Kaye & David A. Freedman, Reference Guide On Statistics, in Reference Manual on Scientific Evidence 351, 378–83 (Federal Judicial Center, 1994))). Such evidence was absent here. Nothing in the record—neither the varying seized amounts, the different sellers, nor the consistency in test results—justifies the inference that a week’s worth of Nos. 22-2994 & 23-1461 13

methamphetamine represented what the conspiracy dealt with during the rest of its thirteen-month run. Given the totality of the circumstances in this case, we are unpersuaded by the fact that the tested amounts represented nearly twenty percent of the total weight of methampheta- mine attributed to the conspiracy. The temporal pitfall per- sists: given the narrow window in which the government col- lected the tested samples, we simply have no idea whether that twenty percent represented the rule or an aberration when it came the purity of the conspiracy’s methampheta- mine. We might have been able to eliminate that speculation had the government tested samples over a greater span of the conspiracy’s existence and come up with consistent purity re- sults. That would be a different case. But in this one, the dis- trict court abused its discretion in finding the week’s worth of samples painted a reliable portrait of the conspiracy as a whole. See United States v. Reed, 72 F.4th 174, 193 (6th Cir. 2023) (finding extrapolation from 2.665 to 4.5 kilograms of actual methamphetamine “inappropriate without sufficient indicia of reliability” (cleaned up)); Titus, 78 F.4th at 600–01 (holding that a sample size of twenty-four out of thousands of illegal prescriptions did not support the district court’s drug quan- tity determination because the government “never showed that the sample was large enough to be reliably representative of the remaining thousands of prescriptions”). Compounding the unreliability of the sample size here, the conspiracy’s source of methamphetamine changed at some point during its operation, and there is no evidence of the purity of the methamphetamine from that new source. To be sure, it appears the supplier changed in the conspiracy’s twilight—around December 2019—but the fact remains that 14 Nos. 22-2994 & 23-1461

the change in supplier prevents us from inferring that con- sistent supply meant consistent purity. See, e.g., United States v. Rodriguez, 666 F.3d 944, 947 (5th Cir. 2012) (finding no error in the court’s inference that unseized drugs “had similar pu- rity levels” where the drugs came from the same supplier at a similar price per weight). Additionally, the evidence showed that Yates purchased methamphetamine from a supplier in Joliet on a weekly basis. As the district court acknowledged, this raises the possibility that the tested samples stemmed from just one purchase from Yates’s supplier, further narrowing the extrapolative utility of the government’s sample. 1 The government points us to our decision in United States v. Castaneda, 906 F.3d 691 (7th Cir. 2018), in which we affirmed the district court’s purity determination even though the gov- ernment tested just one of the six pounds of methampheta- mine attributed to the conspiracy. But we have previously found that Castaneda “fails to provide a clear precedent” for establishing the government’s burden to demonstrate meth- amphetamine purity. Carnell, 972 F.3d at 944. As we explained in Carnell, Castaneda is of little persuasive value in this context because the issue of purity “was not relevant” there: the dis- trict court found that the higher quantity of generic metham- phetamine was the equivalent of the lower quantity of ice. Id. Although “errors in calculating the advisory guideline range are subject to harmless error analysis,“ the district court’s error was not harmless. See United States v. Shelton, 905

1 While the district court made no finding on the issue at sentencing,

it did recognize that the tested samples “could possibly reflect just … one batch” of the conspiracy’s methamphetamine. Nos. 22-2994 & 23-1461 15 F.3d 1026, 1031 (7th Cir. 2018). Had the district court found Yates responsible for just 141.3 grams of actual methamphet- amine (the tested amount), his base offense level would have dropped from 34 to 32, resulting in a final offense level of 29. See U.S.S.G. § 2D1.1(c). That would have yielded a Guidelines range 108–135 months, rather than 135–168 months. The dis- trict court therefore improperly calculated Yates’s Guidelines range, which in this case is a procedural error warranting re- versal. See United States v. Bravo, 26 F.4th 387, 396 (7th Cir. 2022). * * * All this is not to convert drug purity determinations into a math exercise. Reliability remains the “touchstone.” Moore, 52 F.4th at 700. While it would be sufficient for the government to put forward a statistically significant sample, strict statisti- cal validity is not a prerequisite of reliability. We might view things differently had the government tested a similar num- ber of samples that had been seized over a longer period of time, or had it paired those samples with additional reliable evidence that the conspiracy always sold the same metham- phetamine. Alternatively, as other circuits have suggested, the government could have taken judicial guesswork out of the equation by including language in the plea agreement identifying the substances involved. See, e.g., United States v. Roman, 121 F.3d 136, 141 n.4 (3d Cir. 1997). We also continue to leave the door open to the possibility that expert testimony could provide a sufficiently reliable basis for the district court’s extrapolation—for example, had the government put forward a witness testifying that she had never seen metham- phetamine from foreign sources ever test below 99% purity. See Carnell, 972 F.3d at 942–44. 16 Nos. 22-2994 & 23-1461

But that kind of evidence will have to wait until the next case. “When the government fails to meet its burden to sup- port uncharged drug quantities, ‘the government is not per- mitted on remand to try again and submit new evidence in a belated effort to carry its burden.’” United States v. Gibbs, 26 F.4th 760, 767 (7th Cir. 2022) (quoting United States v. Noble, 367 F.3d 681, 682 (7th Cir. 2004)). The district court should re- sentence Yates without any new evidence of drug purity from the government on remand. B. Connelly We turn next to Connelly, who identifies different alleged errors in the district court’s drug quantity calculation. We consider first his contention that the court should not have re- lied on certain coconspirator statements to calculate the total drug weight attributable to the conspiracy. We then turn to his claim that the district court should not have held him re- sponsible for that entire drug weight. 1. Coconspirator Statements “[Q]uantities matter in drug cases.” United States v. Helding, 948 F.3d 864, 866 (7th Cir. 2020). Yet calculating drug quantities is often not as simple as pulling out a scale, nor is it “an exact science.” United States v. Jones, 56 F.4th 455, 506 (7th Cir. 2022) (quoting United States v. Austin, 806 F.3d 425, 431 (7th Cir. 2015)). As this case illustrates, the drugs pur- chased or seized from the defendant and his coconspirators may not represent the full scope of their illegal enterprise. The Guidelines therefore empower district courts to approximate drug quantities based on evidence in the record. U.S.S.G. § 2D1.1 cmt. n.5; see also Austin, 806 F.3d at 431. Such evidence may include, for example, “the price generally obtained for Nos. 22-2994 & 23-1461 17

the controlled substance, financial or other records, similar transactions in controlled substances by the defendant, and the size or capability of any laboratory involved.” U.S.S.G. § 2D1.1 cmt. n.5. Courts may also consider the uncontested findings in a PSR, see United States v. Long, 639 F.3d 293, 299– 300 (7th Cir. 2011), as well as testimony “about the frequency of dealing and the amount dealt over a specified period of time,” United States v. Hernandez, 544 F.3d 743, 746 (7th Cir. 2008). Still, the Guidelines do not allow “nebulous eyeballing.” United States v. Gibson, 996 F.3d 451, 464 (7th Cir. 2021) (quot- ing Hollins, 498 F.3d at 631). District courts must make reason- able, even if imprecise, estimates based on information with sufficient indicia of reliability. United States v. Bozovich, 782 F.3d 814, 818 (7th Cir. 2015); see also United States v. Freeman, 815 F.3d 347, 354 (7th Cir. 2016). Connelly argues that the district court erred by consider- ing Rechkemmer’s and Phelps’s statements in determining the conspiracy’s total drug weight because both had previ- ously lied to law enforcement. Specifically, Rechkemmer ini- tially reported that she received a significantly greater amount of methamphetamine from Yates—as much as six to twelve ounces per week, as opposed to six to twelve ounces in total. She later admitted to being “super high” and “over-ex- aggerat[ing]” during that police interview, and revised her ac- count to the lower estimate. Phelps, one month after admit- ting to receiving two to three ounces of methamphetamine per week, divulged that “she was not initially truthful” in her post-arrest statement because she “minimized [coconspirator Jerel] Guarin’s involvement,” although she did not retract the amount of methamphetamine she claimed to have received. 18 Nos. 22-2994 & 23-1461

The district court did not abuse its discretion by finding both Rechkemmer’s and Phelps’s drug quantity statements reliable. At the outset, we question how “unreliable” Phelps and Rechkemmer really were. While Rechkemmer’s conflict- ing statements related to drug quantity, the PSR and the dis- trict court ultimately adopted her updated and more con- servative estimate. See United States v. Henderson, 58 F.3d 1145, 1152 (7th Cir. 1995) (“We encourage courts to make conserva- tive estimates, especially when presented with generalized testimony, as a way to meet their duties to approximate drug quantities and do so based on trustworthy information.”); see also Bozovich, 782 F.3d at 818. And it appears Phelps’s decep- tion had nothing at all to do with the drug quantities she re- ported, but instead concerned her characterization of Guarin’s involvement. In any event, both statements were consistent with and corroborated by evidence elsewhere in the record. Rechkem- mer and Phelps were not alone in stating that the conspiracy moved at least several ounces of methamphetamine a week. Several other coconspirators—whose credibility Connelly does not question—told law enforcement that Yates was transporting similar quantities. Guarin reported that Yates transported four to eight ounces (113.4 to 226.8 grams) of methamphetamine from Joliet each week. Trenton Sealock stated that Yates would transport five or six ounces (141.75 to 170.1 grams) per week. And John Yates admitted to buying fourteen grams of methamphetamine from Christopher Yates “six or seven” times. The controlled buys and seized drugs further corroborate these amounts. The buys spanned four months, totaled 158.723 grams, ranged from 0.5 to 110.2 grams per buy, and Nos. 22-2994 & 23-1461 19

averaged 15.87 grams. In just a one-week span, the govern- ment purchased more than 143 grams of methamphetamine. These numbers harmonize with Rechkemmer receiving at least 170 grams during the conspiracy, and with Phelps re- ceiving several ounces per week for several months. This is therefore not a case in which the district court esti- mated drug quantity using an “arbitrary” multiplier because it lacked faith in a witness’s account. Henderson, 58 F.3d at 1152. Rechkemmer’s and Phelps’s statements were hardly outliers—the district court appropriately relied on the cocon- spirators’ testimony “in combination with other evidence.” United States v. Acevedo, 28 F.3d 686, 689 (7th Cir. 1994). All that evidence points in one direction: the conspiracy involved at least several ounces of methamphetamine a week. Between the controlled buys and the consistent statements of five dif- ferent coconspirators, the district court did not abuse its dis- cretion in finding Rechkemmer and Phelps reliable. See Free- man, 815 F.3d at 354 (finding no abuse of discretion where the allegedly unreliable testimony “was at least partially corrob- orated by additional evidence drawn from surveillance, trash pulls, seizures, and other witness accounts”); United States v. Valdez, 739 F.3d 1052, 1053–55 (7th Cir. 2014) (same where statements at issue were “consistent and corroborated one an- other”); United States v. Griffin, 806 F.3d 890, 894 (7th Cir. 2015) (same where the district court’s “conservative” estimate was consistent with the admissions of other co-defendants and “information from drugs seizures and controlled buys”). 2. Quantity Attributable to Connelly As to Connelly’s argument that the district court should not have attributed that full weight to him, “[a] defendant in a drug conspiracy is responsible ‘not only for drug quantities 20 Nos. 22-2994 & 23-1461

directly attributable to him but also for amounts involved in transactions by coconspirators that were reasonably foreseea- ble to him.’” Jones, 56 F.4th at 506 (quoting Freeman, 815 F.3d at 353); see also U.S.S.G. § 1B1.3 cmt. n.3. The question here is thus one of reasonable foreseeability, which “refers to the scope of the agreement … entered into when [a defendant] joined the conspiracy, not merely to the drugs he may have known about.” United States v. Flores, 5 F.3d 1070, 1083 (7th Cir. 1993). “A co-conspirator’s conduct is reasonably foresee- able if the defendant-conspirator ‘demonstrated a substantial degree of commitment to the conspiracy’s objectives, either through his words or his conduct.’” United States v. Goodwin, 496 F.3d 636, 642 (7th Cir. 2007) (quoting United States v. Zarnes, 33 F.3d 1454, 1474 (7th Cir. 1994)). Connelly compares himself to a simple street-level dealer who shared a common source of supply with other dealers (Yates) but who otherwise acted independently. See, e.g., U.S.S.G. § 1B1.3(a)(1) cmt. n.4(C)(vi)–(vii). That characteriza- tion undersells his involvement. In fact, Connelly helped hatch the operation. As Rechkemmer told law enforcement af- ter her arrest, Connelly joined coconspirators Guarin, defend- ant Yates, John Yates, and Amanda Edwards to come “up with a plan to distribute methamphetamine to make money to sup- port their own methamphetamine habit.” Connelly supplied one of the coconspirators, Sealock, with methamphetamine to sell on behalf of himself and Christopher Yates. Additionally, he shared a home with coconspirator Edwards, who told law enforcement that Yates had been to her house “ten to fifteen times” with coconspirator Guarin. These kinds of intimate re- lationships and frequent associations support the district court’s finding that the conspiracy’s full drug weight was rea- sonably foreseeable to Connelly. See United States v. Jarrett, 133 Nos. 22-2994 & 23-1461 21 F.3d 519, 531 (7th Cir. 1998) (finding the defendant aware of the scope of the conspiracy’s operations where, among other things, he “dated [a coconspirator] and had a close relation- ship with her son”). Connelly also actively assisted other coconspirators in fur- thering the conspiracy’s goals and avoiding detection. After Guarin’s arrest, Connelly reportedly hid Guarin’s supply of methamphetamine and then returned it once Guarin was out of jail. He also instructed others on how to send money to Yates and permitted Yates to conduct drug transactions out of his home. With all this, the district court had ample support for its conclusion that Connelly sufficiently enmeshed himself in the conspiracy’s operation to find its full drug weight rea- sonably foreseeable to him. III. Conclusion For these reasons, we vacate Yates’s sentence and remand for resentencing consistent with this opinion. We affirm Con- nelly’s sentence.