dissenting.
The majority in my view incorrectly concludes that a conviction under Connecticut General Statute § 53a-179b necessarily qualifies as a “violent felony” under the *95Armed Career Criminal Act, 18 U.S.C. § 924(e)(2)(B) (“ACCA”). Accordingly, I dissent. Although the Connecticut statute is entitled “Rioting at correctional institution,” it criminalizes much more than rioting. The statute sweeps broadly, punishing a range of prison infractions, including passive and nonviolent ones such as disregarding an order to move, engaging in a work stoppage, or counseling another inmate to “disobedience.” Much, if not most, of the behavior penalized by the statute is akin to disorderly conduct or disturbing the peace, neither of which are violent felonies under the ACCA. Such behavior must, of course, be controlled and punished in order to maintain prison discipline, but it is simply not the type of behavior at which the ACCA is directed. The majority’s error could, in my view, have been easily avoided by the simple, painless expedient of applying the modified categorical approach and requiring the government to show what part of the statute Johnson actually violated. This approach is required by our precedent. See United States v. Mills, 570 F.3d 508, 511 (2d Cir.2009) (“When a statute encompasses both violent and non-violent felonies ... we make a limited inquiry into which part of the statute the defendant was convicted of violating.”).
The ACCA provides a mandatory minimum penalty of 15 years of incarceration for a felon who has been convicted of possessing a firearm under 18 U.S.C. § 922(g) and who has three previous convictions for a “violent felony or a serious drug offense.” 18 U.S.C. § 924(e)(1). In the absence of this provision, Johnson’s felon-in-possession conviction would carry a statutory maximum punishment of 10 years. 18 U.S.C. § 924(a)(2). The Sentencing Guidelines also contain a provision that significantly increases a defendant’s offense level if he is subject to an enhanced sentence under the provisions of 18 U.S.C. § 924(e). See U.S.S.G. § 4B1.4. The basic purpose of the ACCA is to combat “the special danger created when a particular type of offender — a violent criminal or drug trafficker — possesses a gun.” Begay v. United States, 553 U.S. 137, 146, 128 S.Ct. 1581, 170 L.Ed.2d 490 (2008). The ACCA accomplishes this goal by looking to a defendant’s past crimes as an indicator of “the kind or degree of danger the offender would pose were he to possess a gun.” Id.
Unless a defendant’s conviction under § 53a-179b is actually for rioting, it is far from obvious that such an offense tends to indicate that he is likely to pose any special danger with a gun. An inmate who, for example, engages in a non-violent strike or counsels a fellow inmate to “disobedience” while incarcerated is simply not involved in the type of conduct with which Congress was concerned when it passed the ACCA. The majority recognizes that “some arguably nonviolent conduct — such as a hunger strike — might violate” § 53a-179b, but nevertheless concludes that a conviction under the statute categorically constitutes a violent felony under the ACCA because it believes that a “ ‘typical instance of this crime’ will indeed involve violent and aggressive conduct.” Maj. Op. at 91 (quoting United States v. Daye, 571 F.3d 225, 234 (2d Cir.2009)). But the majority uses the wrong method of analysis. WTien analyzing a statute like § 53a-179b, which lists a series of different criminal acts, the question is not whether a “typical instance” of the crime involves violent and aggressive conduct, but whether each of the different statutorily delineated criminal acts typically involves such conduct. See Shepard v. United States, 544 U.S. 13, 125 S.Ct. 1254, 161 L.Ed.2d 205 (2005). The answer to that question is “no.”
*96Section 53a-179b provides that “[a] person is guilty of rioting at a correctional institution when he incites, instigates, organizes, connives at, causes, aids, abets, assists or takes part in any disorder, disturbance, strike, riot or other organized disobedience to the rules and regulations of such institution.” Conn. Gen.Stat. § 53a-179b(a). Although the statute’s “nine verbs ... are not nine separate statutorily proscribed methods of violating this statute,” State v. Roque, 190 Conn. 143, 460 A.2d 26, 31 (1983), the statute does proscribe five distinct criminal acts: (1) disorders, (2) disturbances, (3) strikes, (4) riots, and (5) organized disobedience to a prison’s rules and regulations. See State v. Pascucci, 164 Conn. 69, 316 A.2d 750, 752 (1972) (noting that “other organized disobedience” refers to a separate category of conduct, namely, disobedience that is organized). As explained below, besides rioting itself, the four other acts prohibited under § 53a-179b are not “violent felonies” within the ACCA’s definition of that term.
For purposes of the ACCA, a “violent felony” is:
any crime punishable by imprisonment for a term exceeding one year, or any act of juvenile delinquency involving the use or carrying of a firearm, knife, or destructive device that would be punishable by imprisonment for such term if committed by an adult, that-
(i) has as an element the use, attempted use, or threatened use of physical force against the person of another; or
(ii) is burglary, arson, or extortion, involves use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.
18 U.S.C. § 924(e)(2)(B). The district court found that Johnson had committed a “violent felony” pursuant to subsection (ii). Johnson, of course, committed none of the four so-called ‘exemplar crimes’ specifically named in subsection (ii). Instead, the district court concluded that Johnson’s violation of § 53a-179b fell under the ACCA’s so-called “residual clause,” which covers “conduct that presents a serious potential risk of physical injury to another.”
To qualify as a violent felony under the residual clause, a crime must be “roughly similar, in kind as well as in degree of risk posed,” to the ACCA’s exemplar crimes. Begay, 553 U.S. at 143, 128 S.Ct. 1581. Put differently, the applicable test is a two-pronged inquiry, asking (1) whether the crime at issue presents a serious potential risk of physical injury to another, and (2) whether the crime, “in a fashion similar to burglary, arson, extortion, or crimes involving the use of explosives, ‘typically involve[s] purposeful, violent, and aggressive conduct.’ ” United States v. Daye, 571 F.3d 225, 234 (2d Cir.2009) (quoting Begay, 553 U.S. at 144-45, 128 S.Ct. 1581) (other citations omitted). My dissent addresses the second prong of this test.
Section 53a-179b does not define the terms “disorder,” “disturbance,” “strike,” “riot” or “organized disobedience,” but it seems clear to me, by virtue of both commonsense and common parlance, that out of these five terms, “riot” stands apart. “Disorder” is presumably akin to disorderly conduct, a Class C misdemeanor under Connecticut law encompassing acts as nonviolent as “mak[ing] unreasonable noise” and “obstructing] vehicular or pedestrian traffic.” Conn. Gen.Stat. § 53a-182. “Disturbance” is presumably similar to “Creating a public disturbance,” a mere “infraction” under Connecticut law encompassing acts as nonviolent as “annoying] or interfering] with another person by offensive conduct” and “making] unrea*97sonable noise.” Conn. Gen.Stat. § 53a-181a. Strikes and organized disobedience, of course, are not typically crimes outside the confínes of a prison. While I concede that some of these acts may appear relatively more serious in the prison context, context alone does not transform them from offenses that are frequently nonviolent into offenses that are typically purposeful, violent, and aggressive in a fashion similar to burglary, arson, extortion, or crimes involving the use of explosives.1 The fact that a nonviolent, non-aggressive act takes place “in a prison environment” certainly affects the degree of risk posed by that act, but does not, in my opinion, “qualitatively” transform the “kind” of act under consideration. See Maj. Op. at 92. “Riot in the second degree,” on the other hand, has much more in common with the ACCA’s exemplar crimes: it is a Class B misdemeanor under Connecticut law that must involve “tumultuous and violent conduct” engaged in “with two or more other persons.” Conn. Gen.Stat. § 53a-176 (emphasis added).2
The majority errs by failing to apply the modified categorical approach to a statute encompassing, as separately listed acts, these varied, mainly nonviolent, offenses. “When the law under which the defendant has been convicted contains statutory phrases that cover several different generic crimes, some of which require violent force and some of which do not, the modified categorical approach ... permits a court to determine which statutory phrase was the basis for the conviction by consulting the trial record.” Johnson v. United States, — U.S.—, 130 S.Ct. 1265, 1273, 176 L.Ed.2d 1 (2010) (internal quotation marks omitted). More specifically, when, like § 53a-179b, a statute “places together in a single numbered statutory section several different kinds of behavior” that it “separately describes,” but only some of the separately described behavior qualifies as a violent felony under the ACCA, courts must employ the modified categorical approach and treat the ACCA-predicate behavior “as a separate crime” from the nonACCA-predicate behavior. Chambers v. United States, — U.S. —, 129 S.Ct. 687, 691, 172 L.Ed.2d 484 (2009) (applying a modified categorical approach to separate into two crimes a statute criminalizing both an escape from prison and a failure to report to prison).
Accordingly, we should analyze disorders, disturbances, strikes, riots, and other organized disobedience as five separate acts, determining whether each one, standing alone, constitutes a violent felony under the residual clause. Instead, the majority errs by using a “categorical approach,” asking merely whether a conviction under § 53a-179b is violent in the ordinary case, regardless of the underly*98ing facts.3 The categorical approach is generally the proper methodology to use when analyzing statutes. See United States v. Rosa, 507 F.3d 142, 151 (2d Cir.2007). However, where, as here, “the statutory definition of the state crime of conviction encompasses both crimes that would qualify as a ‘violent felony’ and crimes that would not ... we make a limited inquiry into which part of the statute the defendant was convicted of violating.” United States v. Mills, 570 F.3d 508, 511 (2d Cir.2009) (internal quotation marks omitted). Under this modified categorical approach, the government has the burden to prove that the facts underlying the defendant’s conviction show that he was actually convicted under the part of the statute that constitutes a violent felony. Id. at 512.
In this way, the Supreme Court applied the modified categorical approach in Shepard v. United States, 544 U.S. 13, 125 S.Ct. 1254, 161 L.Ed.2d 205 (2005), to analyze a Massachusetts burglary statute which provided that:
Whoever, in the night time, breaks and enters a building, ship, vessel or vehicle, with intent to commit a felony, or who attempts to or does break, burn, blow up or otherwise injures or destroys a safe, vault or other depository of money, bonds or other valuables in any building, *99vehicle or place, with intent to commit a larceny or felony, whether he succeeds or fails in the perpetration of such larceny or felony, shall be punished by imprisonment in the state prison for not more than twenty years or in a jail or house of correction for not more than two and one-half years.
Mass. Gen. Laws Ann., ch. 266, § 16 (West 2008). Shepard “found that the behavior underlying, say, breaking into a building, differs so significantly from the behavior underlying, say, breaking into a vehicle, that for ACCA purposes a sentencing court must treat the two as different crimes.” Chambers, 129 S.Ct. at 690 (emphasis added).
Structurally, § 53a-179b resembles the Massachusetts burglary statute at issue in Shepard in that it criminalizes both violent and nonviolent acts, and lists those acts as separate elements of a disjunctive list. Just as the burglary statute groups burglary of a “building, ship, vessel or vehicle,” § 53a-179b groups a prisoner’s participation in a “disorder, disturbance, strike, riot or other organized disobedience.”4 For the reasons stated above, participation in a riot differs significantly from participation in a strike, disorder, disturbance, or organized disobedience, in that riots are typically violent, whereas these other categories quite frequently may not be. Therefore, just as Shepard treated burglary of a building, but not burglary of a vehicle, as an ACCA predicate, so too should we treat a prisoner’s participation in a riot, but not a prisoner’s participation in the other acts criminalized by § 53a-179b, as an ACCA predicate. Accordingly, the government should be required to prove that Johnson’s conviction under § 53a-179b was actually for rioting.
Ultimately, to be similar in kind to an ACCA exemplar crime, the conduct at issue must be “such that it makes it more likely that an offender, later possessing a gun, will use that gun deliberately to harm a victim.” Begay, 553 U.S. at 145, 128 S.Ct. 1581. Without knowing anything whatever about what Johnson actually did, I do not believe it is possible to conclude that he engaged in conduct suggesting that he poses a special threat with a firearm. To keep order, prisons rightfully require inmates to adhere to strict rules in virtually every aspect of their daily lives. Violations of those rules, such as those criminalized by § 53a-179b, must not go unpunished, but we must also recognize *100that such violations are, unfortunately, common in penal institutions, and often involve conduct entirely dissimilar to the violent and aggressive conduct typified by burglary, arson, extortion, or crimes involving the use of explosives. Before we require defendants like Johnson to spend at least five extra years in prison, beyond any state sentence, based on a prior conviction under § 53a-179b, I would remand and require the government to establish the type of conduct that led to the conviction. Because the stakes for defendants are high, this is a matter of serious concern to judges. In any event, this showing is one required by law.
. Connecticut’s own prison regulations confirm this conclusion. The Connecticut Department of Correction’s definition of "Creating a Disturbance" includes an “inmate work stoppage,” its definition of "Causing a Disruption” includes "harassing others by taunts, name calling or pushing, rattling bars, banging utensils or other objects or in any other manner creating loud or disturbing noises,” and its definition of "Disorderly Conduct” includes “[a]ny nuisance or annoying behavior which interferes with the unit's order.” Connecticut Department of Correction, Code of Penal Discipline, Administrative Directive 9.5, available at http://www.ct.gov/doc/LIB/ doc/PDF/AD/ad0905.pdf (last visited April 28, 2010).
. Rioting in the second degree appears to be the least severe rioting offense under Connecticut law. "Rioting in the first degree” and "Inciting to riot” are both Class A misdemeanors, and also involve "tumultuous and violent conduct.” See Conn. Gen.Stat. § 53a-175; Conn. Gen.Stat. § 53a-178.
. In support of its argument that a violation of § 53a-179b "typically involve[s].... violent, and aggressive conduct,” the majority relies on statistics provided by the government that purport to analyze every reported case that (1) involves a conviction under § 53a-179b, and (2) contains a discussion of the facts underlying the conviction. Maj. Op. at 90. Although statistics may be useful in determining whether a crime is typically violent or not, the government's statistical analysis, on which the majority uncritically relies, is in this instance intensely superficial and incomplete, covering only 11 reported cases across the 38 years since the statute's passage in 1971. Since the government's statistics are limited to reported decisions, they say nothing about how the statute was administered in the vast majority of cases that did not generate published opinions. They do not, for example, indicate the total number of convictions under the statute during that period (although we assume there were far more than 11), or provide any specific account of the factual basis for the convictions they do analyze. As a result, nearly all of the relevant questions about how the statute may have been administered remain unanswered. Why the majority could be reassured by such statistics is not clear. Cf. Chambers v. United States,-U.S.-, 129 S.Ct. 687, 692, 172 L.Ed.2d 484 (2009) (relying, in part, on a rigorous empirical study by the Sentencing Commission showing no violence occurred in 160 cases over a two-year span).
Moreover, the mere 11 cases analyzed by the government include at least one example suggesting that § 53a-179b does not, in fact, typically involve violent and aggressive conduct when a riot is not involved. Most of the cited cases do involve defendants who participated in riots. However, in State v. Rivera, 30 Conn.App. 224, 619 A.2d 1146 (1993), the defendant was convicted under § 53a-179b for refusing to obey a prison guard’s order to stop marching in a circle and to leave an outdoor yard, prior to the outbreak of a riot that he is not alleged to have participated in; no one was injured, and despite the government’s indication that the offense “involved a weapon,” I can find no mention of the defendant having used one. Meanwhile, as Johnson argues, the government's statistics do not account for State v. Pascucci, 164 Conn. 69, 316 A.2d 750 (1972), which involved a defendant who "sat down in a corridor that was meant to be kept clear, refused to move when ordered to do so, waived his arms and loudly urged the inmates to sit down and not to move.” Id. at 752 (construing the statutory predecessor to § 53a-179b). Rivera and Pascucci more than adequately demonstrate "a realistic probability, not a theoretical possibility, that the State would apply its statute to” nonviolent conduct. See James v. United States, 550 U.S. 192, 208, 127 S.Ct. 1586, 167 L.Ed.2d 532 (quoting Gonzales v. Duenas-Alvarez, 549 U.S. 183, 193, 127 S.Ct. 815, 166 L.Ed.2d 683 (2007)); Maj. Op. at 92.
. The majority’s references to United States v. Thrower, 584 F.3d 70 (2d Cir.2009), and United States v. Daye, 571 F.3d 225 (2d Cir.2009), Maj. Op. 91, are therefore inapposite, because neither of the statutes at issue in those cases separately listed the conduct that the respective petitioner alleged was nonviolent, making a modified categorical approach inappropriate. In Thrower, the statute at issue, larceny from the person under New York law, provides: "A person is guilty of grand larceny in the fourth degree when he steals property and when ... [tjhe property, regardless of its nature and value, is taken from the person of another.” N.Y. Penal Law § 155.30(5). Even though the majority is correct that larceny from the person might include pickpocketing, pickpocketing is not separately listed as an offense. Similarly, although the Vermont sexual-assault-of-a-child statute at issue in Daye “encompasses not only forcible assault but also sexual contact to which a child professes to consent,” 571 F.3d at 231, the statutory language does not separately mention consensual and non-consensual contact.
In Daye, we recognized the existence of the modified categorical approach but left open the question of “whether a statute is susceptible to such analysis when it encompasses both violent felonies and other crimes, but does not describe the violent felonies only in distinct subsections or elements of a disjunctive list.” 571 F.3d at 229 n. 4. The clear implication of this observation is that where, as in the instant case, a statute does describe violent felonies and other crimes as elements of a disjunctive list, the statute is susceptible to a modified categorical analysis.