concurring in part and dissenting in part:
I concur in my colleagues’ determination that Officer Dillard was entitled to qualified immunity on Correll Thomas’' claims for unlawful seizure and excessive force under 42 U.S.C. § 1983, and that the district court’s grant of partial summary judgment to Thomas must accordingly be reversed. I write separately, however, because we have repeatedly (and correctly) recognized the unique dangers law enforcement officers face when responding to domestic violence calls — including the inherent volatility of a domestic violence scene, the unique dynamics of battered victims seeking to protect the perpetrators of abuse, the high rate of assaults on officers’ person, and the likelihood that an abuser may be armed. I would therefore hold that the domestic violence nature of the call can alone give rise to reasonable suspicion necessary to justify a Terry frisk.
I also write separately because I do not subscribe to the majority’s holding that Dillard violated Thomas’ Fourth Amendment right by detaining him for purposes' of performing a Terry frisk. See Maj. Op. at 871. My colleagues reach this conclusion by ignoring the axiomatic rule that, at the qualified immunity stage, we review “the objective reasonableness of a particular seizure under the Fourth Amendment .... from the perspective ‘of a reasonable officer on the scene, rather than with the 20/20 vision of hindsight.’ ” Plumhoff v. Rickard, — U.S. -, 134 S.Ct. 2012, 2020, 188 L.Ed.2d 1056 (2014). Here, Officer Dillard responded to not one, but two, domestic violence radio calls, both reported in the span of 33 minutes. The first call reported domestic violence by a black male on the Escondido campus. The second indicated that a man wearing a purple shirt was being observed on a live video feed “pushing” a woman in a secluded part of campus. Officer Dillard encountered a black male wearing a purple shirt — a person matching the description of the suspect of both calls — in the area where at least one of the reported incidents of violence had been observed, with a female who identified herself as Thomas’ girlfriend. These facts supported a reasonable inference' of continuous domestic abuse lasting upwards of half an hour — a situation that the majority (incorrectly, in my view) minimizes as a “simple battery.” See Maj. Op. at 883. True, there is evidence in the record that Thomas’ girlfriend, Amy Husky, claimed that Thomas was just “playing around” when he con-cededly pushed Husky against the storage containers, but a reasonable officer operating without the benefit of hindsight and with the knowledge of the overwhelming frequency with which domestic violence victims seek to protect their abusers, see discussion, infra at pp. 894-95, could have believed himself to be confronting a suspect capable of significant violence who might be armed. By analyzing the circumstances confronting Officer Dillard through a rosy-hued lens of hindsight, the majority opinion deprives future law enforcement officers faced with domestic violence disputes of an important tool for protecting themsélves and the citizens with whose safety they are entrusted. '
I.
It is hornbook law that, at the summary judgment • stage, courts “are required to view the facts and .draw reasonable infer-*893enees ‘in the light most favorable to the party opposing the [summary judgment] motion.’ ” Scott v. Harris, 550 U.S. 372, 378, 127 S.Ct. 1769, 167 L.Ed.2d 686 (2007) (“In qualified .immunity cases, this usually means adopting .... the plaintiffs version of the facts.”). , But it is equally axiomatic that, in determining .whether a police officer is entitled to qualified immunity, we view the facts “from the perspective ‘of a reasonable officer on the scene, rather than with the 20/20 vision of hindsight.’ ” Plumhoff v. Rickard, — U.S. -, 134 S.Ct. 2012, 2020, 188 L.Ed.2d 1056 (2014). My colleagues accept Thomas’ version of the facts, but fail to consider how those facts would have looked to a reasonáble officer on the scene. Properly viewed, the facts confronting Officer Dillard supported his formulation of a reasonable suspicion that Thomas may have been armed.
Officer Dillard received two police dispatches on the afternoon in question. The first reported “domestic violence involving a black male” at the Escondido campus. Approximately thirty minutes later, and within minutes of Officer Dillard’s arrival at the campus, a second police dispatch came in, this time advising Officer Dillard that a male wearing a purple shirt was being observed on live security video pushing a female by some containers in the campus’ biohazard storage area. Thus, both dispatches reported potential domestic violence by a male on a female on the Escondido campus.1 Viewing the facts “from the perspective of a reasonable officer at the scene,” the undisputed similarity between the calls gave rise to a reasonable inference that the reported domestic violence may have been occurring continuously from the time the first call came in through the time of the second (more than thirty minutes later). Officer Dillard, equipped with only this limited information, approached the location where the observed violence had occurred — an alley flanked by large storage containers. Officer'Dillard testified that, given" the nature of the call, the possibility of weapons was the “most important question on [his] mind.” His training had táught.him 'that “a domestic violence call can be. a very dangerous call to a police officer” and that he stood, a good likelihood of “getting seriously hurt qr injured.”
At first, Dillard could not locate the suspect. Only after reversing his car and driving down the alley a second time did Dillard spot Thomas and Husky emerging frpin behind some biohazard storage containers. Thomas, a black male, was wearing a long purple shirt, baggy jeans, skater shoes, a neck chain, an earring, and a black beanie. Thomas therefore matched perfectly the description of both dispatches. Moreover, Dillard found Thomas with Amy Husky, who shortly after encountering Officer Dillard yeíled that Thomas was her “f* * *ing boyfriend” (establishing a' domestic relationship).
Officer Dillard, who stands only five feet, six inches tall and weighs 155 pounds, exited his vehicle and walked approximately four steps, stopping ten feet away from the place where Thomas and Husky were standing. Officer Dillard concedes that he did not observe any visible injuries to Husky’s person. Nor did Amy appear “distraught.” Nevertheless, Dillard had been taught that a domestic violence scene, by its nature, is “very volatile”; people can “suddenly become violent or angry,” making it important to secure the scene.,as quickly as possible, regardless of whether the officer encounters active violence upon arrival.
*894Officer Dillard initiated contact with the couple by telling Thomas and Husky, “No one is in trouble here.” He asked the pair for identification and told them he was “investigating suspicious activity.” Dillard then asked Thomas whether he had any weapons on him, and Thomas said “no.” Officer Dillard asked Thomas if he would “mind” if Dillard nonetheless searched him for weapons — a “common question when investigating a possible domestic violence.” 2 Thomas responded that he “did mind.” Dillard tried to “deescalate” the situation, explaining that Thomas was not under arrest, but that Dillard was “investigating a male wearing a purple, shirt pushing a girl around” and just needed to check Thomas for weapons to make sure tfie scene was safe. Thomas continued to say “no” in response to Officer Dillard’s repeated requests that Thomas consent to a weapons search.
For the next six minutes, Officer Dillard continued trying to elicit cooperation without resorting to force. When verbal commands did not work, Dillard stepped forward in an attempt to put Thomas in a control hold, but Thomas retreated closer to the “recessed” area by the storage containers where the couple had previously been concealed. Officer Dillard “backed off,” afraid that Thomas might try to fight him in the “small, confined space,” and withdrew his taser. Dillard told Thomas to put his hands in the air, to step forward, and to drop to his knees slowly. Thomas said “no.” Dillard repeatedly commanded Thomas to raise his hands, but Thomas refused. Officer Dillard repeatedly commanded Thomas to move his hands away from his waistline — the area Dillard was concerned might conceal a weapon — but Thomas ignored that command, too. After this went on for six minutes, Dillard' deployed his taser, feeling like that was the “final qnd.only option” left to him.
These facts are undisputed. Indeed, the parties-’ recollection as to the encounter differs only slightly. Dillard testified that Husky did not say -that the pair had been kissing until after the incident, at which point she also admitted that Thomas had pushed her. Thomas, on the other hand, declares that Husky told Dillard within the first moments of the encounter that the pair had “just [been] kissing” behind the containers. Thomas also recalls that Husky started “yelling” at some point during the stand-off that Thomas had “done nothing wrong.” Finally, Thomas insists that at no time during the encounter did he act “belligerently]” or “aggressively]” toward Officer Dillard or anyone else.
But even crediting Thomas’ version of events, we must consider how a reasonable officer would perceive this scene, armed only with the limited information available to Officer Dillard. Officer Dillard encountered an individual suspected of significant domestic violence, who “appeared to be hiding behind the biohazard storage area” (a relatively secluded area of campus where even Thomas admits the couple had gone “for privacy”). Thomas and Husky appeared “startled” by Officer Dillard’s presence. Officer Dillard perceived Thomas as “standoffish,” in part because Thomas spoke in a low, serious tone. Contrary to my colleagues’ suggestion, neither Thomas’ statement that he was not “belligerent” nor any other evidence in the record contradicts Officer Dillard’s perception of Thomas’ manner. The majority opinion points out that, according to Thomas, Officer Dillard appeared “surprised” and then “angry and agitated” after Thomas refused to consent to a weapons search. But such *895a reaction would be a natural response to a confrontation with a suspect an officer reasonably suspects to be armed, who is refusing to permit the officer to conduct a weapons search.
It is also undisputed that Thomas refused to move his hands from his waistline, despite numerous requests. Thomas characterizes his hand placement as “normal if I was standing and making casual conversation with someone.” But in light of Dillard’s repeated requests that Thomas move his hands away■ from his waistline, a reasonable officer in Officer Dillard’s position could well have perceived the natural “fidgeting” of Thomas’ hands by his sides as an attempt to conceal a weapon. Of course, we know now that Thomas was not concealing a weapon in his waistband. But Officer Dillard — who reasonably believed he was dealing with a demonstrably violent, domestic violence suspect — had , no way of knowing that.
The majority opinion makes much of Husky’s statement to Dillard that the couple had just been “kissing.” But it is well-documented that victims of domestic violence — even those who initially report their abusers to police — more often than not “recant or refuse to cooperate” with the police seeking to help them. See, e.g., Tom Lininger, Prosecuting Batterers After Crawford, 91 Va. L.Rev. 747, 751 (2005) (“Approximately 80 percent of victims decline to assist the government in prosecutions of domestic violence cases.”); Lisa Marie De Sanctis, Bridging the Gap Between the Rules of Evidence and Justice for Victims of Domestic Violence, 8 Yale J.L. & Feminism 359, 367-68 (1996) (“[Vjictims of domestic violence are uncooperative in approximately eighty to ninety percent of cases. Many victims are uncooperative from the initial filing of the case — ”). Thus, even assuming that Husky was yelling at Officer Dillard that Thomas had done nothing wrong, a reasonable officer in Dillard’s position could have discounted those statements as an effort by the victim of violence to protect her abuser. Thus, contrary to the majority’s suggestion, the resolution of this factual discrepancy in Thomas’ favor does not negate Dillard’s reasonable belief that Thomas was a perpetrator of potentially severe and/or continuous domestic violence. Only with the benefit of hindsight do we know otherwise. Under these circumstances, Dillard reasonably believed that he had a duty, as a community caretaker, not to leave a domestic violence victim in the care of a demonstrably violent person without conducting a weapons search.
II.
Thomas argues that the Fourth Amendment precluded Dillard from frisking Thomas for weapons' under the circumstances presented here. Thomas' additionally argues that this conclusion was clearly established as of September 2010, and thus Officer Dillard was not entitled to qualified immunity. The qualified immunity inquiry is two-fold: It requires a violation of a constitutional right, and a finding that such right was clearly established at the time the violation occurred. Pearson v. Callahan, 555 U.S. 223, 231, 129 S.Ct. 808, 172 L.Ed.2d 565 (2009).3 I agree with the majority’s holding that it was not clearly established in September 2010 that Dillard’s conduct was unconstitutional. However, I disagree with the majority’s holding that Thomas has demonstrated a violation of his Fourth Amendment right, because I would find that a reasonable officer in Dillard’s position could have for*896mulated a reasonable suspicion that Thomas might be armed based on the facts available to, Dillard here. -
In determining whether Dillard’s frisk violated a constitutional 'right, this court should bear in mind that a Terry frisk is a “protective search — permitted without 'a warrant and on the basis "of reasonable suspicion.” Minnesota v. Dickerson, 508 U.S. 366, 373, 113 S.Ct. 2130, 124 L.Ed.2d 334 (1993). Unlike a full-blown arrest, “[t]he protective search for weapons .•.. constitutes a brief, though far from inconsiderable, intrusion upon the sanctity, of the person.” Terry v. Ohio, 392 U.S. 1, 26, 88 S.Ct. 1868, 20 L.Ed.2d 889 (1968). “[Bjeasonable suspicion” that a person is armed is enough — and this is “less than probable cause.” Minnesota, 508 U.S. at 373, 113 S.Ct. 2130. In other words, we must merely find that under “the facts available, to the officer at that moment” a “man. of reasonable caution” “might” believe that a person was “armed and presently . dangerous.” Pennsylvania v. Mimms, 434 U.S. 106, 111-12, 98 S.Ct. 330, 54 L.Ed.2d 331 (1977) (internal quotation marks omitted); see also Gallegos v. City of Los Angeles, 308 F.3d 987, 990 (9th Cir.2002) (“The reasonable suspicion standard ‘is a, less demanding standard than probable cause,’ and merely requires ‘a minimal level of objective justification.’” (quoting Illinois v. Wardlow, 528 U.S. 119, 123, 120 S.Ct. 673, 145 L.Ed.2d 570 (2000))).
The majority today holds that, “although the domestic violence nature of a police investigation is a relevant consideration in assessing whether there is reason to believe a suspect is armed and dangerous, it is not alone sufficient to establish "reasonable suspicion.” Maj. Op. at 871. I disagree.
It is well-established that the nature of a suspected crime can alone support a reasonable suspicion that a suspect might be armed — even where the officer do'es not observe any “physical indication” of a weapon. See, e.g., United States v. Post, 607 F.2d 847, 851 (9th Cir.1979) (“[Djespite the fact that he had not observed a weapon or any physical indication of a weapon, it was reasonable to assume, from the nature of the offense contemplated [drug trafficking], that [the defendant] was armed and dangerous.” (citing Terry, 392 U.S. at 28, 88 S.Ct. 1868)). Indeed, courts have routinely" upheld a right to frisk on the basis that the officer has legitimately stopped a person suspected of a crime, like domestic violence, that is frequently committed with the use of weapons. See, e.g., Terry, 392 U.S. at 28, 88 S.Ct. 1868 (finding it reasonable to assume that men whose behavior suggests they are ■ casing a building for a “daytime robbery” might be armed because of the nature of the crime); United States v. Mattarolo, 209 F.3d 1153, 1158 (9th Cir.2000) (Terry frisk of nighttime burglary suspect was constitutional); People v. Shackelford, 37 Colo.App. 317, 546 P.2d 964, 966-67 (1976) (police could lawfully frisk an individual who was - found in the immediate vicinity' of the area from which a rape victim had fled because the individual “matched the description of a suspect who had allegedly committed an act of violence” — rape).
Like burglary and robbery, domestic violence is frequently committed with a firearm or other weapon. For example, 36.7 percent of female domestic violence shelter residents in California reported having at some point been threatened or harmed with a firearm during a domestic dispute. See Susan B. Sorenson, Ph.D., & Douglas J. Wiebe, Ph.D., Weapons in the Lives of Battered Women, 94 Am. J. Pub. Health 1413 (2004). Another study by the Department of Justice found that roughly 15 percent of all incidents of domestic violence involve a weapon. See Callie Marie Rennison, Ph.D., U.S. Dep’t of Justice, Re*897port No. NCJ 187635, Bureau of Justice Statistics Special Report: Intimate Partner Violence and Age of Victim (1993-99) 7 (Oct.2001). And in 2011, domestic violence assaults- involving a firearm accounted for nearly two-thirds of all fatal, shootings of female victims. See Violence Pol’y. Ctr., When Men Murder Women: An Analysis of 2011 Homicide Data 6 (Sept.2013).
Indeed, the majority concedes that some domestic violence calls can be “dangerous,” even deadly, for all parties involved. Maj. Op. at 880. However, the majority emphasizes that domestic violence encompasses a broad spectrum of “criminal acts of varying degrees of seriousness” to support its holding that the domestic violence nature of a call cannot justify a Terry frisk. Maj. Op. at 885. But the majority’s argument ignores the practical reality that an officer often does not know what part of that spectrum of criminal acts he will confront when he arrives at the scene of a domestic dispute. Indeed, this is what distinguishes domestic violence from other “broad categories” of crimes — like theft. Whereas there is little risk that an officer investigating a suspected mail thief will instead encounter an armed bank robber, cf. United States v. Flatter, 456 F.3d 1154, 1158 (9th Cir.2006) (cited by the majority) (“Mail theft by postal employees is not a crime that is frequently associated with weapons ..”), an officer investigating a report of “simple battery” may well find that the domestic violence has escalated to assault with a deadly weapon by the time the police arrive.4
The majority also reiterates that Thomas was observed committing only a “simple battery” (referring to the second dispatch). See, e.g., Maj. Op. at 883. But in fact, there were two reports of domestic violence, and Thomas matched’the description of both reports. The first report did not specify the precise behavior it characterized as “domestic violence” — it could well have involved more severe abuse, including with a weapon.5 And in any event, even a “simple battery” can quickly escalate into something far more dangerous. Unless an officer has credible evidence that a domestic violence suspect is «of'armed, a reasonably cautious officer would assume that a domestic violence suspect “might” be armed — and that is all that is required here. The majority’s analysis is thus im-permissibly infused with the benefit of hindsight.
Further undermining the majority’s view is our circuit’s repeated observation that “[t]he volatility of situations involving domestic violence” makes them particularly dangerous. Marquez v. City of Phoenix, 693 F.3d 1167, 1175 (9th Cir.2012), as amended on denial of reh’g (Oct. 4, 2012) (alteration in original) (quoting Mottos v. Agarano, 661 F.3d 433, 450 (9th Cir.2011)). We have also said that “[w]hen officers respond to a domestic abuse call, they understand that ‘violence may be lurking *898and explode with little warning.’ ” United States v. Martinez, 406 F.3d 1160, 1164 (9th Cir.2004) (quoting Fletcher v. Clinton, 196 F.3d 41, 50 (1st Cir.1999)) (“[M]ore officers are killed or injured on domestic violence calls than on any other type of call.” (quoting Hearings before Senate Judiciary Committee, 1994 WL 530624 (F.D.C.H.) (Sept. 13, 1994) (statement on behalf of National Task Force on Domestic Violence))); see also Hopkins v. Bonvicino, 573 F.3d 752, 766 (9th Cir.2009) (stating the same).
Consistent with our precedent, Officer Dillard’s training had taught him that “a domestic violence call can be a very dangerous call to a police officer.” The undisputed evidence in the record (a declaration by a police procedures expert) indicates that 31.7 percent of assaults on. police officers in 2011 occurred “while answering domestic violence type radio calls,” and 12.7 percent of the 72 officers killed in 2011 were answering domestic violence calls (citing the Federal Bureau of Investigation’s (“FBI’s”) 2011 annual publication on police assaults and deaths).
My colleagues in the majority are correct that the evidence in the record appears somewhat to overstate the FBI data it purports to summarize, because the figures cited in fact encompass assaults and deaths attributable to all kinds of “disturbance calls” — not just domestic violence calls. See Maj. Op. at 881 n. 11. But the majority ignores that the FBI data nevertheless supports our circuit’s oft-stated conclusion that domestic violence is frequently associated with weapons and is one of the more dangerous calls an officer can receive! For example, the FBI’s 2011 statistics show that approximately 33 percent of the assaults on police officers in 2011 (18,216 out of 54,774 assaults) were committed while police were responding to “disturbance call[s],” a category which includes domestic violence calls. FBI, Law Enforcement Officers Assaulted: Circumstances at Scene of Incident by Type of Weapon and Percent Distribution, 2011 (last visited Feb. 17, 2016), https://www.fbi. gov/about-us/cjis/ucr/leoka/2011/tables/ table-73 [hereinafter “Table 73”]. And of these 18,216 assaults, 17.3 percent (3,155 assaults annually) were committed with a weapon. Id.
Indeed, there is reason to think that domestic violence calls are more dangerous to police officers than calls relating to ■other crimes which we have, already deemed to give rise, by their very nature, to a reasonable suspicion that a suspect might be armed. In United States v. Mattarolo, 209 F.3d 1153 (9th Cir.2000), for example, we held that a Terry frisk of a nighttime burglary suspect was constitutional, based on the dangerous nature of the crime. Id. at 1158. But the FBI’s data shows that two of the 72 police officers killed in 2011 were responding to a “domestic disturbance,” while zero officers were killed in 2011 while responding to a “burglary in progress” or the “pursuing a burglary suspect.” See FBI, Law Enforcement Officers Feloniously Killed: Circumstance at Scene of Incident, 2002-2011 (last visited Feb. 17, 2016), available at https://www.fbi.gov/about-us/cjis/ucr/ leoka/2011/tables/table-19 [hereinafter “Table 19”]. In fact, from 2002 to 2011, more than three times as many police officers were killed while responding to domestic violence calls than while responding to burglary calls. Id. (showing that 36 officers were “feloniously killed” while responding to domestic violence calls, as compared to only 11 officers responding to active burglary calls). And in 2004, 16 percent (9 out of 57) of the police officers feloniously killed in the line of duty were responding to domestic disturbance calls, specifically. Id. ■
*899Instead of looking at this hard data or following our own precedent, however, the majority credits a recent article and the findings of one researcher whose study was limited to the Los Angeles area, which suggest that robbery and burglary calls are actually more dangerous than domestic violence calls. See Maj. Op. at 880-81 & n. 11.6 Even assuming, arguendo, that robbery and burglary calls are more frequently associated with weapons than domestic violence calls, it does not follow that domestic violence calls do not also pose a substantial threat to a police officer’s safety. In holding that a “daylight robbery .... would be likely to involve the use of weapons,” Terry made no suggestion that robbery represented the outer limit of crimes thought to be sufficiently dangerous to support a reasonable inference on the part of the officer that a person sus-' pected of that crime might be armed. See Terry v. Ohio, 392 U.S. 1, 28, 88 S.Ct. 1868, 20 L.Ed.2d 889 (1968). Rather, Terry recognized that even where an officer does not have the probable cause necessary to make an arrest, it would “be clearly unreasonable to deny the officer the power to take necessary measures to determine whether the person is in fact carrying a weapon and to neutralize the threat of physical harm,” as long as the “officer is justified in believing that the individual whose suspicious behavior he is investigating at close range is armed and presently dangerous to the officer or to others.” Id. at 24, 88 S.Ct. 1868.
The FBI data referenced in the record shows that- thousands of police officers are assaulted every year while responding to domestic violence disputes. One in every five of these assaults involves a deadly weapon (a firearm, a knife, or another “dangerous weapon”). See Table 73. And every year, these assaults result in deaths to law" enforcement officers. See Table 19. Given these statistics, I would find that domestic violence calls are associated with weapons frequently enough that an officer can form a reasonable suspicion that a suspect might be armed based on the domestic violence nature of the call.7
Of course, an officer’s reasonable suspicion could in some circumstances be dispelled by other facts known to the officer. See Terry, 392 U.S. at 28, 88 S.Ct. 1868 (suggesting that, under certain circumstances, a suspect’s response to an officer’s approach might dispel reasonable suspicion that suspect is armed). But at the time he encountered Thomas, Dillard was aware of no such facts. Dillard encountered an individual matching the descrip*900tion of a suspect of two reports of domestic abuse, who emerged “suspiciously” from a concealed area where the violence had allegedly occurred -with a woman. Thomas was wearing baggy clothes and a long shirt that could conceal weapon. Thomas acted “standoffish.” He refused to consent to a weapons search. More importantly, he refused even to raise his hands or to move them away from his waistline, where Dillard was reasonably concerned a weapon might be concealed.8
True, Thomas’ baggy clothing could not alone support a reasonable suspicion that he was armed. See Ybarra v. Illinois, 444 U.S. 85, 93, 100 S.Ct. 338, 62 L.Ed.2d 238 (1979). (government failed to “articulate any specific fact that would have justified” a reasonable suspicion that Ybarra was armed and dangerous when “the most” the officer “could point to was that Ybarra was wearing a 3/4-length lumber jacket, clothing which the State admits could have been expected on almost any tavern patron” at that time of year). But where an officer is confronting the suspect of a violent crime frequently committed with weapons, the fact that the suspect’s clothing could conceal a weapon is relevant to determining whether reasonable suspicion that the 'suspect might be armed is dispelled.9
The majority focuses on the lack of active violence when Officer Dillard arrived. But even if the initial violence has subsided, it is not uncommon for the perpetrator, or even the victim, to erupt into'sudden violence or anger at an officer responding to a domestic violence call. See Martinez, 406 F.3d at 1164; cf. Lininger and Marie De Sanctis, supra. If that 'occurs, the seemingly calm scene can turn dangerous or deadly (if weapons are present) in a matter of seconds. Given this volatility, a reasonably cautious police officer will start by securing the domestic violence scene to ensure there are no weapons on hand if someone does become violent — exactly what Officer Dillard tried to do here. In fact, an officer has a duty as a “community caretak[er]” to take reasonable steps to protect the. victim of reported domestic violence, and Officer Dillard’s “failure to do so would have amounted to a dereliction of duty.” United States v. Willis, 431 F.3d 709, 713 (9th Cir.2005).
Perhaps with the benefit of “20/20 ... hindsight” we might say that Officer Dillard should have realized that Thomas was not armed. But in the Fourth Amendment context we judge the reasonableness of an officer’s conduct “from the perspective of a reasonable officer on the scene.” Graham v. Connor, 490 U.S. 386, 396, 109 S.Ct. 1865, 104 L.Ed.2d 443 (1989) (citing Terry, 392 U.S. at 20-22, 88 S.Ct. 1868 (1968)); cf. Mattarolo, 209 F.3d at 1158 (“The officer must choose between being sure that the suspect is not armed and jeopardizing his own safety.”).
In my view, the nature of a domestic violence call justifies an officer’s formulation of a reasonable suspicion that a suspect may be armed (in the absence of mitigating circumstances). Indeed, a law enforcemént officer should feel a moral duty to protect victims of domestic vio*901lence. And while an officer of course cannot shadow an abuser forever to ensure no future violence befalls the victim, he can at least make sure that the “scene is secure” from possible violence with weapons before he leaves the victim with an abuser. Given that domestic violence assaults account for nearly two-thirds of all fatal shootings of female victims, in the United States, see Violence Pol’y Ctr. Report, supra, I would find that the relatively small incursion on bodily dignity of permitting a Terry frisk based on the domestic violence nature of a police call is far outweighed by the strong law enforcement interest — indeed, the broader societal interest — in saving the lives of domestic violence victims. 'We achieve this interest by assuring that abusers are not armed before the police officer leaves. the scene of a domestic violence encounter. Yet the majority today requires a law enforcement officer to face potential liability unless he leaves a domestic violence scene without any assurance that the abuser is not armed and will not again inflict violence on the victim — only next time, with a gun. I cannot endorse a legal rule that places law enforcement officers in such a stressful dilemma. Thus, while I agree with the result reached by the majority today, I cannot join my colleagues to the extent they deny law enforcement officers a tool — the Terry frisk — vital for ensuring their own safety, as well as the safety of the domestic violence victims they have a duty to protect.
. Despite believing he had a right to conduct a Terry frisk, Officer Dillard asked Thomas for consent in efforts to avoid a physical altercation and to resolve the investigation peacefully.
. Courts of appeal have discretion to address either prong first. Pearson, 555 U.S. at 236, 129 S.Ct. 808.
. Given the demonstrated danger to police officers associated with domestic violence calls, the majority’s citation to cases involving non-violent crimes not associated with weapons misses the point. See Maj. Op. at 876, 878-79, 887-88 (citing, e.g., United States v. Flatter, 456 F.3d 1154, 1158 (9th Cir.2006); Ramirez v. City of Buena Park, 560 F.3d 1012, 1016 (9th Cir.2009) (suspicion that a person found sleeping in his car was under the influence of drugs did not give rise to a reasonable suspicion that suspect was armed)). As the majority recognizes, see Maj. Op. at 880, domestic violence' can be very "dangerousf]” and can involve crimes clearly associated with weapons — like assault- with a deadly weapon or homicide. Thus domestic violence is of a fundamentally different character than the crimes that we háve found do not justify a Terry frisk.
. Because Thomas matched .the description of both reports, and because both reports involved domestic violence occurring on the Escondido campus, Dillard reasonably believed the dispatches were related.
. For example, the majority credits the opinion of the author of an article appearing in the May 2011 edition of The Police Chief, which (like the majority) editorializes on the FBI’s statistics presented herein, but does not actually present any contrary evidence to support its conclusion. In any event, I find it odd that the majority chooses to credit this article over the well-supported conclusions reached by at least two Ninth Circuit cases. See, e.g., Hopkins v. Bonvicino, 573 F.3d 752 (9th Cir.2009); United States v. Martinez, 406 F.3d 1160 (9th Cir.2004), discussed supra, at 897.
. The warrantless entry cases .cited by the majority for the proposition that "domestic abuse cases [do not] create a per se exigent need for warrantless entry,” do not resolve whether a frisk — which, unlike warrantless entry, is necessitated by safety concerns — may be justified by the domestic violence nature of the call. See Maj. Op. at 882 (citing United States v. Black, 482 F.3d 1035 (9th Cir.2007), and United States v. Martinez, 406 F.3d 1160 (9th Cir.2005)). Moreover, both cases cited by the majority upheld the constitutionality of the officers’ warrantless entry in response to a domestic violence call. See Black, 482 F.3d at 1040-41; Martinez, 406 F.3d at 1165. Thus, these cases hardly support' the majority’s conclusion that Officer Dillard violated Thomas’ Fourth Amendment rights in conducting a Terry frisk of an individual matching'the description of two reports of domestic violence on the Escondido campus.
. The mere fact that Thomas provided answers to Officer Dillard’s questions (i.e. refusals to cooperate) did not "negate” Dillard’s "reasonable hypothesis” that he might be armed. See Terry, 392 U.S. at 5, 28, 88 S.Ct. 1868 (defendants mumble in response to officer s question did not negate a reasonable hypothesis that the defendant might be armed where the officer had observed suspicious activity that led him to believe the defendant intended to commit-a crime involving weapons).-
_ por example, reasonable suspicion that a suspect was armed might be dispelled if the suspect’s clothing made clear he would have nowhere t0 ^ a weapon.