Case Information
*1 Before B RENNAN , S CUDDER , and S T . E VE , Circuit Judges . S CUDDER , Circuit Judge
. On an afternoon in December 2012, the Chicago Police Department received an anonymous 911 call reporting a Hispanic man in a black sweater and black hat, carrying a bag, and climbing under a warehouse fence. O cers arrived and found someone who matched the de- scription, but after stopping and frisking him, determined he was not engaged in any crime. The initial suspect then pointed the o cers to someone else nearby who was crossing *2 the street and walking toward the police. This man, Anthony Howell, was white and wearing a black jacket and dark hat. When an o ffi cer approached to ask what was going on, How- ell did not answer, looked panicked, and put his hands in his pockets. The o ffi cer reacted by pa tt ing down Howell and found a gun in his jacket. A federal gun charge followed, and Howell moved to suppress the gun as the fruit of an uncon- stitutional stop-and-frisk. The district court denied the mo- tion, Howell proceeded to trial, and a jury found him guilty.
Howell now appeals from the denial of the suppression motion. In evaluating his position, we also confront a question about the proper scope of the record on review. The question is whether we limit our review to the pretrial record or ex- pand our look to consider the arresting o cer’s trial testi- mony as well. The answer ma ers because the facts in the pre- trial record di ered in a material way from those that emerged at trial, where the arresting o cer testi ed that he decided to proceed with the pat down only after Howell ig- nored a directive to remove his hands from his pockets. In the end, we limit ourselves to the pretrial record, for that is the only source of facts the district court considered in denying Howell’s motion. Viewing that record as a whole, we con- clude that police lacked reasonable suspicion to frisk Howell. We therefore reverse the denial of his suppression motion and vacate his conviction for possessing that gun.
Our reversal is only partial, however, because Howell was also convicted on a second gun charge. Three months af- ter the December 2012 stop-and-frisk, police executed a war- rant to search Howell’s apartment, where they found more guns and ammunition. There was ample evidence for the *3 jury to fi nd that Howell possessed the guns in his apartment, so we a ffi rm his conviction for this separate o ense.
I
A Around noon on December 4, 2012, an anonymous 911 caller reported that a Hispanic man wearing a black sweater, black hat, and black bag was climbing under a fence at a ware- house on South Artesian Avenue in Chicago’s Brighton Park neighborhood. O ffi cers Sean Kelly and Christopher Miller ar- rived about fi ve minutes later and saw a man, Eric Escobar, who matched the caller’s description and was walking on the sidewalk outside the warehouse. The o ffi cers stopped Escobar and immediately pa ed him down but found nothing suspi- cious. Escobar explained that he worked at the warehouse and had stepped outside to buy a drink and a snack for his manager. The manager emerged from the building and con- fi rmed that account while also verifying Escobar’s identity.
While talking to police, Escobar noticed another person nearby—a white man wearing a black jacket and a dark hat who was walking toward the o cers. He was later identi ed as Anthony Howell, who lived across the street. Upon rst noticing Howell, Escobar remarked that he seemed to match the police’s account of the 911 caller’s description. O cer Kelly reacted to Escobar’s comment by calling out to Howell from across the street and asking, “What’s going on?” Accord- ing to Kelly, Howell refused to answer and instead did a “quick double take,” had “a look of panic on his face,” and placed his hands in his pockets. Finding this reaction suspi- cious, Kelly approached Howell and immediately frisked him for weapons.
As soon as O ffi cer Kelly began the frisk, he felt a hard ob- ject in Howell’s jacket pocket. When asked what it was, How- ell replied, “protection.” When Kelly tried to retrieve the gun, Howell pulled away, started to run, but quickly slipped on gravel and fell. At some point in the ensuing scu ffl e, a .38 cal- iber Smith & Wesson revolver fell out of Howell’s pocket, and the police secured it and placed Howell under arrest.
B
A federal grand jury later charged Howell with unlaw- fully possessing a gun as a prior convicted felon, a violation of 18 U.S.C. § 922(g)(1). Before trial Howell moved to suppress the gun, arguing that the police violated his Fourth Amend- ment rights by stopping and frisking him without reasonably suspecting him of being engaged in criminal activity. Howell also sought an evidentiary hearing on the motion.
The district court denied both requests in an oral ruling. It rst denied Howell’s request for a hearing on the ground that he had not shown a material factual dispute. While Howell’s brief in support of his motion contested the o ffi cers’ version of events—for example, he denied refusing to answer O ffi cer Kelly’s question—the court emphasized that Howell had stopped short of submi ing an a ffi davit swearing under oath to the same representations. Without such an a ffi davit, the district court reasoned, Howell failed to create a genuine fac- tual dispute that warranted a hearing.
From there the district court relied on police paperwork and FBI reports of interviews with O cer Kelly and the other o cers involved in the stop-and-frisk to rule on the merits of Howell’s Fourth Amendment challenge. Applying the famil- iar reasonable suspicion standard from Terry v. Ohio , 392 U.S. *5 1 (1968), the district court considered the totality of the cir- cumstances and began by observing that Howell came close enough to matching the 911 caller’s description to authorize the stop—while not Hispanic, he was wearing a black jacket and a dark hat. The district court also emphasized that How- ell reacted to O cer Kelly’s question about what was going on by refusing to answer, doing a double take, looking pan- icked, and pu tt ing his hands in his pockets. The combination of these reactions and circumstances, the court concluded, not only supplied the reasonable suspicion necessary to support O cer Kelly’s stop of Howell, but also suggested that he may have been concealing something—thereby authorizing the pat down.
Howell proceeded to a jury trial, where he renewed his motion to suppress at the close of evidence. The district court made quick work of the renewed motion, observing that the ma tt er had been fully resolved pretrial. The court therefore denied the motion “for the reasons that [were] already given in addressing the defendant’s prior motions on the same topic.” Neither party said a word about any aspect of the trial evidence a ecting or informing the court’s prior ruling.
The jury returned a guilty verdict. Howell then moved for a new trial or a judgment of acqui al, again arguing that the district court should have granted his prior motion to sup- press. The district court construed Howell’s argument as re- newing the motion to suppress for a third time. And the court reacted by referring to its pretrial ruling—reiterating that it had “already ruled on [the] Fourth Amendment issue”—and “incorporate[d] by reference everything that [it had] already said.” Throughout this colloquy nobody referred to the trial evidence.
Howell now appeals.
II
A Before reaching the merits, we confront an interesting and challenging question about how to de fi ne the scope of the rec- ord on appeal when reviewing a motion to suppress. While preparing for oral argument, we noticed a material di ff erence between the facts relevant to the suppression motion in the pretrial record and those elicited at trial. Speci fi cally, at trial, O cer Kelly testi fi ed that he had “asked [Howell] to take his hands out of his pocket, and [Howell] didn’t respond to that, either.” But the pretrial record on which the district court based its denial of Howell’s motion made no mention of Of- cer Kelly giving such a direction or Howell then ignoring it. More to it, the police paperwork and FBI reports (prepared in anticipation of federal charges being brought against Howell) only referenced Howell pu ing his hands in his pockets.
The di ff erence may be signi cant. If true, the additional fact that Howell disobeyed a police order to remove his hands from his pockets would have bolstered suspicions that he was armed and dangerous.
It is less clear, however, that we should consider trial testi- mony. The district court did not. It denied Howell’s motion to suppress (and post-trial renewal of the motion) entirely—100 percent—on the pretrial record (the police and FBI paper- work), giving no e ect whatsoever to any aspect of O cer Kelly’s trial testimony.
It warrants emphasis that the district court acted well within its discretion in handling Howell’s motions in this way. When a defendant chooses to renew a suppression *7 motion at or after trial, a district court is free to incorporate its past reasoning, as it did here, or alternatively to consider evi- dence introduced at trial. Unless the parties bring new evi- dence gleaned from trial to the court’s a ention, the law does not compel either approach, and we leave the choice to the district court. Relatedly, the path the district court chooses may inform our own discretion as to which facts warrant con- sideration on appeal. See United States v. Hicks , 978 F.2d 722, 724–25 (D.C. Cir. 1992) (employing similar reasoning).
Recognizing the importance of these questions and con- siderations to our analysis, we asked the parties for supple- mental briefs addressing the proper scope of the record.
B
Ordinarily we de fi ne the record on appeal by limiting the facts we review to those considered by the district court. De- fi ning the appellate record in this way ensures that we per- form our role as a court of review while also respecting the district court’s role as fact fi nder. See Midwest Fence Corp. v. U.S. Dep’t of Transp. , 840 F.3d 932, 946 (7th Cir. 2016) (“As a general rule, we will not consider evidence on appeal that was not before the district court when it rendered its decision.”); see also W RIGHT & M ILLER , F EDERAL P RACTICE & P ROCEDURE § 3956.1 (5th ed.) (noting that “[t]he court of appeals’ concern with the scope of the record stems in part from the notion of the respective roles of the appellate and trial courts,” espe- cially “the district court’s role as the rst-instance nder of fact”).
But here the government presses a di erent approach, in- viting us to consider both pretrial and trial evidence. No other appeal has required us to confront this question in much *8 depth, and what case law does exist sends di ering signals. The government’s position nds support in cases like United States v. Parra , where we a ffi rmed the denial of a defendant’s pretrial motion to suppress cocaine found in a search incident to arrest. 402 F.3d 752, 767 (7th Cir. 2005). In identifying the standard of review, we observed—without saying more— that “[i]n reviewing a denial of a suppression motion, we may consider evidence introduced both at the suppression hearing and at trial.” Id. at 764; see also United States v. Duguay , 93 F.3d 346, 350 (7th Cir. 1996) (stating without further elaboration that we may consider evidence presented at trial in reviewing a district court’s pretrial ruling denying a motion to sup- press).
On the other side of the ledger come cases like United States v. Smith , 80 F.3d 215 (7th Cir. 1996). The district court there denied the defendant’s pretrial motion to suppress ma- rijuana found in his car during a tra ffi c stop. See id. at 218–19. On appeal the defendant pointed to evidence elicited at trial to argue that no reasonable o cer would have made the stop. See id. at 220. But we declined to consider the trial evidence, instead basing our review “solely on what the district court knew at the time of the ruling.” Id. ; see also United States v. Fryer , 974 F.2d 813, 819 (7th Cir. 1992) (limiting our review of a suppression ruling to the pretrial record in the same way).
To be sure, even decisions like Parra and Smith , while lim- iting review to the pretrial record, say very li le (if anything) about why that approach is proper. Perhaps our most fulsome treatment of the issue came in United States v. Longmire , 761 F.2d 411 (7th Cir. 1985). There we took guidance from the Su- preme Court’s 1925 decision in Carroll v. United States , where the Court relied in part on trial evidence to a rm the denial *9 of a suppression motion. See 267 U.S. 132, 162 (1925). Follow- ing suit in Longmire , we observed that we had the discretion to consider trial evidence in reviewing a pretrial suppression ruling. See 761 F.2d at 418 (“[E]vidence adduced only at trial may be used to sustain the denial of a motion to suppress.”).
We took care, however, to sound caution in the exercise of that discretion, recognizing that consideration of the trial tes- timony presents both bene ts and risks. See id. On the one hand, using that testimony “avoids a windfall reversal of the defendant’s conviction” where the trial record reveals that a police action was constitutional. Id. But just as importantly, “several problems may be presented by such use of trial testi- mony,” including “[p]rejudice to the accused.” Id. A balance must be struck between these two interests—avoiding a windfall reversal of a conviction while also steering clear of unfair prejudice to the defendant.
Our decision in Longmire o ered guidance for achieving that balance. In the ordinary course, we may consider trial tes- timony in reviewing a pretrial suppression ruling. See id. We opted to follow that baseline rule on the facts of Longmire be- cause Darlene Longmire did not contest the trial testimony, let alone a empt to show that it prejudiced her. See id. at 420– 21 (“Longmire apparently believed that [the trial] testimony did not alter the correctness of the pretrial suppression ruling for she failed to ask the trial court to reconsider that ruling.”). Still, we recognized that consideration of the trial testimony would be inappropriate where a defendant shows that doing so would result in prejudice—for example, where “the credi- bility and veracity of a relevant government witness have been put into question by defense counsel” and that witness introduces new facts at trial. Id. at 418.
Considered collectively our prior cases show that the question presented—when we may consider trial evidence in reviewing a pretrial motion to suppress—does not lend itself to bright-line answers. Rather, we approach the inquiry on a case-by-case basis, taking account of all available information regarding the proceedings below. We read our case law (and the principles underpinning it) to at least establish that we re- tain the discretion to consider trial evidence bearing on a dis- trict court’s ruling on a motion to suppress where that evi- dence came into play in the district court’s consideration of the motion—where the defendant renewed the motion and thereby invited the district court to reevaluate its prior ruling in light of trial evidence or where the district court undertook such a reevaluation of its own accord. See Smith , 80 F.3d at 220 (limiting review to the pretrial record in part because the de- fendants did not renew their suppression motions at trial); see also Hicks , 978 F.2d at 725 (observing that when trial evidence casts doubt on a pretrial suppression ruling, the parties should “bring alleged errors to the trial court’s a ention by making a proper objection or ling a motion”). The prior cases likewise counsel that one factor properly informing our exer- cise of discretion is whether considering the trial evidence would cause unfair prejudice to the defendant.
On balance we conclude that these principles tilt against consideration of the trial record here. Foremost, the district court itself never considered O cer Kelly’s trial testimony— neither in denying Howell’s pretrial motion, nor in denying the later renewals of the same motion. To the contrary, the district court made plain that it was denying the renewed mo- tion for the same reasons given in the pretrial ruling. The dis- trict court never hinted that the trial evidence was even *11 relevant, much less that it in any way a ff ected any dimension of the court’s prior reasoning.
So, too, is it clear that it would prejudice Howell to con- sider O cer Kelly’s trial testimony, as it contained a new, ma- terial representation—that Howell disregarded a clear direc- tion from O cer Kelly to remove his hands from his pockets. At trial Howell had no reason to believe the district court would consider that testimony as part of revisiting its pretrial ruling. Even more, Howell may have had sound strategic rea- sons, when cross-examining and a tt empting to impeach Of- fi cer Kelly, not to draw a tt ention to the new fact o ff ered for the fi rst time at trial. See Longmire, 761 F.2d at 418 (emphasiz- ing this precise caution). Howell’s focus at trial was not on somehow seeking to relitigate the motion to suppress before the jury. That Howell renewed the motion during and after trial almost certainly re fl ected nothing more than an e ff ort to ensure preservation of the issue for appellate review.
The upshot is that Howell had li le incentive at trial to fo- cus on factual details pertinent to a pretrial motion that the district court resolved before trial even began. Put another way, Howell had every reason to believe the trial would be all and only about whether the government presented evidence to prove beyond a reasonable doubt that he possessed a gun following a prior felony conviction.
We fi nd it equally noteworthy that neither party’s brie fi ng on appeal even identi ed the di erence between the pretrial and trial records. Both parties addressed the district court’s denial of Howell’s motion to suppress by focusing strictly on the pretrial record. Only after we raised the question at oral argument and requested supplemental brie ng did the *12 government seek to defend the district court’s ruling by rely- ing on O cer Kelly’s trial testimony.
The law does not compel us to consider trial evidence in reviewing a suppression ruling; it merely a ords us the dis- cretion to do so. See Parra , 402 F.3d at 764; Longmire , 761 F.2d at 418. Under these circumstances, we decline to consider Of- cer Kelly’s trial testimony. Doing so would prejudice Howell in a material and unfair way. We therefore look only to the pretrial record in evaluating Howell’s motion to suppress.
III
A A seizure occurs within the meaning of the Fourth Amendment if, in the totality of the circumstances, a reasona- ble person would not feel free to disregard the police and move along . See Florida v. Bostick , 501 U.S. 429, 434 (1991). Here the parties agree that at some point during O cer Kelly’s approach and questioning, Howell was seized. But they dispute whether the seizure was constitutional.
Under the Fourth Amendment, police may stop a person only if they have reasonable suspicion that he is engaged in criminal activity. See Terry , 392 U.S. at 21–22. Our focus on reasonableness “balanc[es] the need to search (or seize) against the invasion which the search (or seizure) entails .” Id. at 21. The inquiry is fact-intensive: we look to the totality of the circumstances to see whether police “ha[d] a particular- ized and objective basis for suspecting the particular person stopped of criminal activity.” United States v. Cortez , 449 U.S. 411, 417–18 (1981); see also United States v. Street , 917 F.3d 586, 593 (7th Cir. 2019) (explaining that reasonable suspicion must *13 be based on speci c, articulable facts that would justify an in- trusion on the suspect’s liberty and dignity).
The district court determined that O ffi cer Kelly’s decision to stop Howell respected the Fourth Amendment. We agree. The police were responding to a 911 call reporting suspicious activity across the street from where Howell was walking. He roughly matched the caller’s description. Howell then reacted to that approach in a way O ffi cer Kelly reasonably could have found suspicious—by doing a double take, taking on a pan- icked look, and refusing to respond. These facts and circum- stances combined to give O cer Kelly su ffi cient reason to ap- proach Howell, and in posing a question to him, to conduct an investigatory stop. We move, then, to the frisk.
B
A frisk—a limited pat down of the suspect’s outer clothing to search for weapons—is permissible under the Fourth Amendment only if a police o cer can “point to speci c and articulable facts” indicating “that criminal activity may be afoot and that the persons with whom he is dealing may be armed and presently dangerous.” Terry , 392 U.S. at 21, 24–25, 30. It is precisely because a frisk is more intrusive than a stop that the Fourth Amendment compels this additional armed- and-dangerous inquiry. See id. at 27.
We begin where the police did—with the information from the 911 call. In assessing a stop-and-frisk based on a tip, we must assess the reliability of the information conveyed by the caller. See United States v. Lopez , 907 F.3d 472, 479 (7th Cir. 2018). Sometimes callers identify themselves and their doing so lends meaningful credibility to the information they pro- vide. Other times callers remain anonymous and reliability *14 comes from independent sources corroborating the tipster’s account. See id. at 480. Along these lines, we have identi fi ed “a spectrum of knowledge and reliability that a ff ects the rea- sonableness of police action taken pursuant to the tip”—at one end, “a tip from a known, trusted, and reliable source,” and at the other, “an anonymous tip without signs of reliabil- ity.” Id. at 479–80. “Tips that come from more trustworthy sources will require less independent corroboration than those obtained from more questionable sources.” Id. at 480.
The call here was anonymous. The Supreme Court has long recognized that “[a]n anonymous tip alone seldom demonstrates the informant’s basis of knowledge or veracity” because that basis is “by hypothesis largely unknown, and unknowable.” Alabama v. White , 496 U.S. 325, 329 (1990) (in- ternal quotations omi ed). Additionally, the tip itself con- tained no further indicia of the informant’s reliability. It also o ered nothing but a barebones description of the suspect: the caller identi ed the suspect’s race (Hispanic), sex (male), and dress (black sweater, hat, and bag). What limited details the caller did supply fell short of describing Howell with sig- ni cant accuracy. Above all else, the call mentioned a bag, which Howell did not have, and described Howell, who is white, as appearing Hispanic.
To justify a frisk of Howell, then, police needed some ad- ditional indicia of reliability or another source of corrobora- tion beyond the limited information provided by an anony- mous caller. But there was none. There were no additional calls to police, for example, or bystanders at the scene who said they had witnessed anything troubling. Remember that the police, upon arriving at the warehouse, spoke to both Eric Escobar and his manager—both of whom worked there and *15 neither of whom expressed any concern about an a empted burglary or other crime. Nor did police arrive to nd any ev- idence of illegal activity themselves.
Consider, too, the nature of the reported o ff ense. A call to police is less likely to support reasonable suspicion in the Terry analysis when it does not describe an ongoing crime or emergency. That precept follows directly from Terry itself. See 392 U.S. at 30 (holding that a frisk is justi ed where there is reasonable suspicion that both the suspect “may be armed and presently dangerous” and “criminal activity may be afoot”); see also Lopez , 907 F.3d at 485 (“[I]nvestigative stops related to completed crimes must be distinguished from in- vestigative stops related to ongoing or imminent crimes.”). We recently emphasized much the same point in United States v. Watson , 900 F.3d 892 (7th Cir. 2018). There police received an anonymous call reporting that “boys” were “playing with guns” by a “gray and greenish Charger” in a parking lot. Id. at 893. Police went to the parking lot, saw a car matching the description, searched it, and found a gun. Id. at 894. We held that the police lacked reasonable suspicion to justify that search, however, because the caller was anonymous and “did not describe a likely emergency or crime.” Id. at 893.
In the same vein, the anonymous tip here merely reported someone climbing a warehouse fence. Nothing about it sug- gested that an emergency was underway or that anybody was in imminent danger. Not a word was said about weapons, an injured victim, or anyone being threatened. The alleged of- fense took place around noon—in broad daylight—and the record is devoid of any evidence that it took place in a high- crime area. By any reasonable measure, the 911 caller de- scribed a low-end, nonviolent o ense—something that surely *16 warranted a police response, but by no means could be con- sidered an emergency. See United States v. Goodwin , 449 F.3d 766, 769 (7th Cir. 2006) (interpreting the Supreme Court’s Fourth Amendment jurisprudence to apply a sliding scale ap- proach where “the amount of permissible intrusion is a func- tion . . . of the gravity of the crime being investigated”); see also United States v. Williams , 731 F.3d 678, 686 (7th Cir. 2013) (explaining that a frisk is a “more burdensome intrusion” on the suspect’s liberty and dignity than a stop alone).
The government sees things di ff erently. It contends that although the district court characterized the o ff ense as a mere “trespass,” a reasonable police o cer could have interpreted the caller to be reporting a felony burglary. And if the re- ported o ense was a felony, the government continues, it was serious enough to justify a frisk, even though the crime was no longer underway by the time police arrived. See United States v. Hensley , 469 U.S. 221, 233–34 (1985) (holding that po- lice had reasonable suspicion to stop a suspect in a felony armed robbery, even though the crime was already complete and therefore no longer in progress).
The government did not raise this argument in the district court, and we are reluctant to entertain a position that Howell had no opportunity to contest. See In re Veluchamy , 879 F.3d 808, 821 (7th Cir. 2018). Regardless, we need not decide whether the conduct described by the 911 caller would con- stitute a misdemeanor or a felony to assess its seriousness. Nor do we expect police to categorize the tips they receive into misdemeanors or felonies before assessing the appropri- ate response to each situation. Sometimes the answer would be easy (shots red and victim screaming); other times it may be next to impossible (suspicious person wearing stocking *17 cap, yelling, and parked in front of neighbor’s house). Requir- ing the police to place the calls they receive into a felony, mis- demeanor, or some other bucket strikes us as unworkable. See United States v. Jones , 953 F.3d 433, 436–37 (6th Cir. 2020) (ex- plaining that the federal courts have avoided a brightline rule for when police may investigate a completed nonfelony, in part because of “the elusive and evolving nature of the felony- misdemeanor distinction”). The be tt er approach is to consider the totality of the information supplied by the 911 caller, in- cluding the nature of the reported crime, the reasonable infer- ences following from the caller’s information, and how the police responded.
Here what ma tt ers perhaps most is that the 911 call in no way suggested that the suspect was armed or dangerous. The caller did not so much as hint at violence, injuries, or weap- ons. Nor did such a threat arise after police responded to the call. The o ffi cers saw no crime in progress and encountered no victim or witnesses. To the contrary, they arrived to nd a rather innocuous scene: Eric Escobar walking on the sidewalk outside a warehouse in broad daylight, on his way to grab snacks for his manager. Indeed, it was Escobar—not any of- cer—who saw Howell and suggested that perhaps he was the one the police were looking for. Then and only then did O cer Kelly turn his a ention to Howell—who presented only as white, not Hispanic as the 911 caller described, and who was not carrying a bag of any kind. And even then, all that O cer Kelly reported seeing from across the street was Howell appear nervous and panicked, fall silent, and put his hands in his pockets. There were no suggestions that Howell was armed—nobody claims he had a bulge in his pocket or made any move to hide anything.
Finally, the government urges us to place substantial weight on Howell’s panicked look upon seeing police. We agree that this nervousness is relevant, but it must be consid- ered against the full context of the circumstances facing the police. Nervousness alone, at least not as a categorical ma tt er, does not create reasonable suspicion that a suspect is armed and dangerous. See Williams , 731 F.3d at 687 (recognizing that “[m]ost people, when confronted by a police o ffi cer, are likely to act nervous, avoid eye contact, and even potentially shift their bodies as if to move away from the area”). Nor does a suspect’s mere refusal to answer an o ffi cer’s questions, with- out more, create reasonable suspicion. See Illinois v. Wardlow , 528 U.S. 119, 125 (2000) (“[A]ny refusal to cooperate, without more, does not furnish the minimal level of objective justi ca- tion needed for a detention or seizure.”).
Nervousness is more salient to the reasonable determina- tion calculus when it accompanies other suspicious behavior or circumstances suggesting a risk to o cer safety. In Ward- low , for example, the Supreme Court held that police had rea- sonable suspicion to stop and frisk a person who, while car- rying a bag and upon making eye contact with the o cers, suddenly started running away. See id. at 122. The suspect’s “[h]eadlong fl ight,” the Court concluded, was “the consum- mate act of evasion,” made all the more suspicious by the fact that his spontaneous bolt occurred in a high-crime area. Id. at 124.
Our case law follows a similar pa ern. In United States v. Brown , we concluded that police had reasonable suspicion to frisk a suspect who was stopped for speeding and then asked to exit the vehicle for a pat down. 188 F.3d 860, 865 (7th Cir. 1999). Not only did the suspect show “excessive nervousness” *19 when police stopped him, but several other facts contributed to reasonable suspicion: his car was under FBI surveillance for possible involvement in a large-scale drug operation, it reeked of marijuana, and he was stopped “in a high crime area where there had been drug activity, shootings, and gang violence.” Id.
More recently, in United States v. Adair , we held that police had reasonable suspicion to stop and frisk a suspect who tried to evade an o ffi cer by weaving through a crowd away from him. 925 F.3d 931, 933 (7th Cir. 2019). Critical to our conclu- sion, however, were the circumstances in which that hap- pened: a woman had reported to police that a group of people she did not recognize were standing outside her apartment smoking, drinking, and engaged in “very suspicious activ- ity.” Id. She described the suspect in speci c terms and stated that he had a black gun in his front pocket. See id. When police arrived, an o ffi cer found a matching suspect with a bulge in his pocket who was trying to evade detection. Id. at 933, 937. This all transpired “late at night in a high-crime area.” Id. at 936. Considering these facts in combination, we held that po- lice reasonably suspected that the suspect was armed and dangerous. See id. at 936–37.
Howell’s case presents far di erent facts and circum- stances. His panicked look and silence in response to O ffi cer Kelly’s question were not accompanied by any a empt to fl ee or any furtive movement. He responded to seeing O cer Kelly as many might, by appearing to want to move along and avoid a discussion with the police. What most concerns us is how O cer Kelly reacted. He did not respond by continuing to approach Howell or allowing more time for further ques- tions—“Where do you live?”, “Do you know anything about *20 a burglary here?”, “Were you trying to climb under this fence?” and the like—but instead by immediately commenc- ing a pat down. Put another way, O cer Kelly reacted to see- ing Howell much like he did to observing Eric Escobar upon arriving at the scene—by seeing that he matched aspects of the caller’s description then instantly pa tt ing him down. But Terry teaches that frisks need to account for the totality of cir- cumstances—they cannot be rote or re fl exive—and here the circumstances required more before O cer Kelly’s encounter with Howell would permit a frisk.
The caution the Supreme Court sounded in Florida v. J.L., 529 U.S. 266 (2000), warrants underscoring. There an anony- mous 911 caller reported that a young, black male wearing a plaid shirt at a bus stop was carrying a concealed gun. See id. at 271. Upon responding to the bus stop, the police saw some- one who perfectly matched that description, pa ed him down, and discovered a gun. See id. The Court held the frisk unconstitutional, because the tip was anonymous and de- scribed only “a subject’s readily observable location and ap- pearance.” Id. at 272. The Court explained that such a descrip- tion is reliable only in that it helps police to identify the ac- cused. See id. Reasonable suspicion required more: the tip needed to be “reliable in its assertion of illegality, not just in its tendency to identify a determinate person.” Id.
These same concerns weigh on us here. The Court found that the tip in J.L. was too barebones to support a frisk because it identi ed only “readily observable” traits such as race, sex, clothing, location, and age. Id. Here the caller provided even less information, all of which was readily observable: race, sex, clothing, and location. And unlike in J.L. , the match was not exact. If the tip in J.L. did not create reasonable suspicion *21 to support a frisk, we fi nd it hard to reach another conclusion here, especially considering the nature of the reported crime—someone climbing under a warehouse fence in broad daylight—and what the police encountered upon arriving, which was nothing suggesting any sort of burglary or tres- pass, much less a violent crime or anyone threatened or in- jured.
We are mindful that police, in carrying out their duties, often must react to potential threats quickly and under di - cult and uncertain circumstances. But having considered the entirety of the facts and circumstances here, we conclude that police did not have reasonable suspicion to frisk Howell. We therefore reverse the district court’s denial of his motion to suppress.
IV
A A nal issue remains for resolution. Following Howell’s initial encounter with police in December 2012, law enforce- ment obtained a warrant and arrested him in his home a few months later. In the course of the arrest, law enforcement con- ducted a protective sweep of Howell’s bedroom and found .32 caliber ammunition hidden inside a sock. The police then ob- tained a warrant to search the entire apartment.
The search revealed a few notable items. First, the police found a North American Arms .22 caliber revolver at the bot- tom of the apartment building’s internal air shaft—an open space just outside of Howell’s window. Police later traced that rearm to Howell’s father, Thomas Howell, who bought it in 1989 and passed away in 2010. Second, the police found a Frontier Derringer .22 caliber revolver, with Thomas Howell’s *22 name engraved on the handle, in a small safe in the living room. The safe also contained .22 and .32 caliber ammunition and a holster bearing a North American Arms logo.
The federal charges brought against Howell came in two counts (both alleging violations of 18 U.S.C. § 922(g)(1)) and covered a total of three guns. Count one addressed the gun the police found in his pocket in December 2012, and count two covered the two additional guns they recovered from his apartment in March 2013. Howell asks us to vacate his convic- tions on both counts. In doing so, he does not argue that the items recovered during the March 2013 search of his apart- ment were the fruit of the unconstitutional pat down that took place in December 2012. He contends in a much less direct way that the jury’s learning about the gun the police found in his pocket during that pat down impermissibly tainted its consideration of the evidence presented on count two.
In support of this argument, Howell points to a few events that unfolded at trial. He testi fi ed in his own defense on count two, stating that he did not know about the guns found in his apartment in March 2013. He emphasized that both guns were stored out of view and were traceable to his deceased father. On cross-examination, however, the government also asked Howell about the evidence against him on count one— the gun found in his pocket in December 2012. Howell refused to answer. He now insists that this refusal damaged his cred- ibility in front of the jury—an outcome he would have avoided if the rearm on count one was never admi ed.
Howell also points to a second event from his trial. Before returning a verdict, the jury posed two questions to the court: rst, whether Howell was arrested in the room overlooking the air shaft where the police found the North American Arms *23 revolver, and, second, where other personal e ff ects of Thomas Howell (a photograph and a memorial CD recovered by po- lice) were stored inside the apartment. The district court de- clined to answer, informing the jury that the evidence was closed and thus that they needed to decide the case on the record as it stood. Howell now argues that the jury’s ques- tions revealed concern about whether he constructively pos- sessed the guns in his apartment, given that they were out of view and traceable to his father.
B
A constitutional error requires reversal unless it was harmless beyond a reasonable doubt—“that is, [unless] no reasonable doubt exists that the error a ff ected the jury’s ver- dict.” United States v. McKinney , 954 F.2d 471, 475 (7th Cir. 1992) (citing Chapman v. California , 386 U.S. 18, 24 (1967)). Where evidence was erroneously admi tt ed at trial, we con- sider several factors in determining whether the evidence was harmless: the prejudicial e ff ect of the evidence, how the gov- ernment used the evidence at trial, the strength of the govern- ment’s case outside that evidence, and “if there are any such indications from the verdict, how the jury likely received and considered the impermissible evidence.” United States v. Mil- ler , 673 F.3d 688, 701 (7th Cir. 2012).
No doubt the wrongfully admi ed gun was harmless as to count two. The government had robust evidence supporting that count. The jury heard about two revolvers tucked away in di ff erent corners of Howell’s apartment. It also heard that police found .32 caliber ammunition stu ed inside a sock near where Howell was standing moments before his arrest—am- munition that matched the .32 caliber North American Arms pistol found in the air shaft just outside his window.
The district court also instructed the jury to consider each count separately. Even if the jury’s questions evinced some degree of reservation as to whether Howell constructively possessed the guns found in his apartment, we have no reason to think that the jury returned a guilty verdict on count two because of the evidence it heard on count one. The govern- ment presented ample independent evidence to support the jury’s verdict on count two. Right to it, the admission of the gun that served as the basis for count one, however wrongful, was harmless beyond a reasonable doubt as to count two.
* * *
The .38 Smith & Wesson revolver recovered from Howell’s pocket in December 2012 should have been suppressed as the fruit of an unconstitutional frisk. Any error in admi ing that gun, however, was harmless as to Howell’s conviction on count two. We therefore REVERSE the denial of Howell’s mo- tion to suppress the gun recovered in December 2012, VACATE his conviction on count one, and AFFIRM his con- viction on count two.