Vamsidhar Vurimindi v. Attorney General United States

PRECEDENTIAL

UNITED STATES COURT OF APPEALS
FOR THE THIRD CIRCUIT
_____________

Nos. 19-1848 & 19-2904
______________

VAMSIDHAR VURIMINDI,
Petitioner

v.

ATTORNEY GENERAL
UNITED STATES OF AMERICA
______________

On Petition for Review of a Decision
of the Board of Immigration Appeals
(Agency No. A096-689-764)

Immigration Judge: Walter A. Durling
______________

Argued: October 27, 2021
______________

Before: GREENAWAY, JR., KRAUSE, and PHIPPS,
Circuit Judges.

(Opinion Filed: August 24, 2022)
Rachel A.H. Horton [ARGUED]
Courtney G. Saleski
DLA Piper
1650 Market Street
One Liberty Place, Suite 5000
Philadelphia, PA 19103
Counsel for Petitioner

Victoria M. Braga [ARGUED]
United States Department of Justice
Office of Immigration Litigation
P.O. Box 878
Ben Franklin Station
Washington, DC 20044
Counsel for Respondent

_____________

OPINION
______________

KRAUSE, Circuit Judge.

We are called on here to decide whether Pennsylvania’s
stalking statute, 18 Pa. Stat. and Cons. Stat. § 2709.1(a)(1),
constitutes a removable offense under the Immigration and
Nationality Act, or, applying the so-called “categorical
approach,” whether the elements of the Pennsylvania offense
are a categorical match to the elements of the generic “crime
of stalking” for which a noncitizen is removable under 8 U.S.C.
§ 1227(a)(2)(E)(i). Because the Board of Immigration Appeals

2
mistakenly found that Petitioner Vamsidhar Vurimindi failed
to raise this issue before the Immigration Judge and denied two
motions for relief on that basis, we must also decide whether
this question is one we may address in the first instance. We
conclude that we can and that because the Pennsylvania
stalking offense sweeps more broadly than the federal generic
under the categorical approach, it is not a removable offense.
Accordingly, we will grant Vurimindi’s consolidated Petitions
for Review.

I. Factual and Procedural Background

Vurimindi, a native of India, came to the United States
on a work visa in 2000, and after marrying an American
citizen, became a lawful permanent resident in 2008. JA 74.
Vurimindi’s erratic behavior towards some of his neighbors
eventually led to his arrest and conviction on two counts of
misdemeanor stalking under Pennsylvania law. JA 74, 187,
244, 420. In relevant part, the Pennsylvania stalking statute
makes it a crime to:

engage[] in a course of conduct or
repeatedly commit[] acts toward
another person ... under
circumstances which demonstrate
either an intent to place such other
person in reasonable fear of bodily
injury or to cause substantial
emotional distress to such other
person[.]

18 Pa. Stat. and Cons. Stat. § 2709.1(a)(1). Vurimindi was
sentenced to two consecutive terms of fifteen to thirty months’

3
imprisonment followed by a period of supervised release. JA
74, 187.

In 2016, while Vurimindi was serving this sentence, the
Government initiated removal proceedings against him under
8 U.S.C. § 1227(a)(2)(E)(i), which makes any noncitizen
convicted of a “crime of stalking” removable from the United
States. JA 72-74. But “crime of stalking” is not defined in the
INA, so to ascertain whether Vurimindi’s Pennsylvania
conviction qualified under this removal provision, the IJ was
required to apply the categorical approach, i.e., comparing the
elements of the relevant state offense with the elements of the
federal generic offense. See Descamps v. United States, 570
U.S. 254, 257 (2013), JA 418-19.

After comparing the elements of a crime of stalking
under the INA to those of the Pennsylvania stalking statute, the
IJ concluded that “[Vurimindi’s] conviction under the
Pennsylvania statute is the prototypical case for stalking as set
forth in [the] INA,” and that Vurimindi was removable on that
basis. JA 419. Vurimindi appealed to the BIA, arguing that
the IJ erred in finding him removable because his “conviction
under 18 PA CSA § 2709.1 do[es]n’t constitute a crime of
‘Stalking’ under INA § 237(a)(2)(E)(i).” JA 496; see JA 492-
96.1

1
The IJ also rejected Vurimindi’s argument that his
convictions were not final, reasoning that Vurimindi’s pending
state case sought post-conviction relief and was not a direct
appeal. JA 418. The BIA likewise rejected Vurimindi’s
challenge to the finality of his conviction. JA 62–63. That
issue is not before us on appeal.

4
 On appeal, the BIA issued three orders, the second and
third of which are before us today. In the first order, the BIA
mistakenly found that Vurimindi “does not contest his
removability” and affirmed without addressing Vurimindi’s
argument that his Pennsylvania conviction was not a
categorical match with the generic offense under the INA.
JA 63 n.4. But see JA 496 (raising challenge to removability
on categorical approach grounds). In the second, the BIA
denied Vurimindi’s motion to reopen, which it construed as a
motion to reconsider the same categorical approach argument
that it had deemed “waived” in the first order and that it
continued to assert had not been “raised earlier in the
proceedings.” JA 68–69. Vurimindi then filed a motion for
reconsideration of that order, which the BIA denied, stating
that there was no “factual or legal error in our [prior] decision
or any aspect of the respondent’s case that was overlooked.”
JA 71.

Currently before us are Vurimindi’s petitions for review
of the second and third of these orders—the denials of his
motion to reopen and motion for reconsideration. JA 11–13.

II. Jurisdiction and Standard of Review

The BIA had jurisdiction over Vurimindi’s appeal
pursuant to 8 C.F.R. §§ 1003.1 and 1240.15, and over his
motion to reconsider under 8 C.F.R. § 1003.2. This Court’s
jurisdiction is governed by 8 U.S.C. § 1252, which provides
for judicial review of final orders of removal. Our review of
the purely legal question presented by this appeal is plenary.
Moreno v. Att’y Gen., 887 F.3d 160, 163 (3d Cir. 2018).

We give deference to the BIA’s definition of a crime of
stalking so long as it is “based on a permissible construction of

5
the statute.” Mondragon-Gonzalez v. Att’y Gen., 884 F.3d 155,
158 (3d Cir. 2018) (citation omitted)). But we do not afford
deference to its categorical approach determination or its
“parsing of the elements of the underlying [state] crime,” Mahn
v. Att’y Gen., 767 F.3d 170, 173 (3d Cir. 2014), because that
analysis is based on interpretation of state law, see Salmoran
v. Att’y Gen., 909 F.3d 73, 77–78 (3d Cir. 2018).

III. Discussion

On appeal, Vurimindi again urges that his Pennsylvania
conviction is not a categorical match to the removable “crime
of stalking” offense under the INA because the Pennsylvania
statute is indivisible, with a single mens rea element that is
satisfied by “either an intent to place [the victim] in reasonable
fear of bodily injury or to cause substantial emotional distress
to [the victim],” 18 Pa. Stat. and Cons. Stat. § 2709.1(a)(1),
while the generic offense in the INA is limited to the intent to
place the victim in “fear of bodily injury or death,” Matter of
Sanchez-Lopez (“Sanchez-Lopez II”), 27 I. & N. Dec. 256, 258
(BIA 2018) (construing 8 U.S.C. § 1227(a)(2)(E)(i)). For its
part, the Government concedes that Vurimindi is not
removable if the state statute is indivisible because that offense
would then sweep more broadly than the INA’s stalking
offense. Oral Arg. Tr. 24:50–25:34; 45:42–59. It contends,
however, that the state statute is properly considered divisible,
with two alternative mens rea elements, one of which—the
“intent to place [the] victim in reasonable fear of bodily
injury”—is a categorical match to § 1227(a)(2)(E)(i).
Answering Br. 18.

So the merits of Vurimindi’s petitions come down to the
divisibility of Section 2709.1(a)(1), but before we can address
that issue, we must determine if we should remand for the BIA

6
to consider that question in the first instance. That is because,
although the parties agree that the BIA erred in holding
Vurimindi waived his challenge to removability, see
Answering Br. 21–22, the BIA never addressed the merits of
the IJ’s conclusion that the statutes were a categorical match,
see JA 61–71. Thus, we must first determine whether remand
is required before we can turn to the proper reading of the
Pennsylvania statute and the proper application of the
categorical approach.

A. Whether Remand to the BIA Is Required

When faced with an issue that the BIA has not yet
addressed, we typically follow the Supreme Court’s instruction
in INS v. Orlando Ventura to “remand a case to an agency for
decision of a matter that statutes place primarily in agency
hands.” 537 U.S. 12, 16 (2002) (per curiam). In that case,
where the Ninth Circuit reversed the BIA’s denial of asylum
based in part on de novo consideration of a changed-
circumstances argument not addressed by the BIA, the
Supreme Court held that failure to follow the ordinary remand
rule violated “basic considerations” of administrative law:
namely that, for a “decision of a matter that statutes place
primarily in agency hands,” remand permits an agency to
“bring its expertise to bear upon the matter; . . . evaluate the
evidence; . . . make an initial determination; and, in doing so,
[the agency] can, through informed discussion and analysis,
help a court later determine whether its decision exceeds the
leeway that the law provides.” Id. at 16–17.

Since Ventura, however, we and other Courts of
Appeals have identified a number of exceptions to the remand
rule where such “basic considerations” of administrative law
do not apply. In Jean-Louis v. Attorney General, for example,

7
we declined to remand for the BIA to apply the categorical
approach and simply applied it ourselves where the BIA had
already utilized its expertise to define the mens rea required for
the generic crime under the INA and the only remaining
issue—application of the categorical approach—did not
implicate the agency’s expertise or statutory domain. 582 F.3d
462, 466–68 (3d Cir. 2009). More recently, in Singh v.
Attorney General, we again declined to remand after
concluding that the BIA erred in opting to apply the categorical
approach instead of the modified categorical approach, and we
applied the modified categorical approach ourselves,
explaining that we would, in any event, be “review[ing] [the
BIA’s determination] de novo as it implicates a purely legal
question.” 839 F.3d 273, 282 (3d Cir. 2016) (quoting Restrepo
v. Att’y Gen., 617 F.3d 787, 790 (3d Cir. 2010)). And the Ninth
and Eleventh Circuits have taken a similar approach. See, e.g.,
Talamantes-Enriquez v. Att’y Gen., 12 F.4th 1340, 1348–49
(11th Cir. 2021); Mandujano-Real v. Mukasey, 526 F.3d 585,
588–89 (9th Cir. 2008).

What these cases teach is that remand to the BIA is not
required and the Court of Appeals may address an issue in the
first instance where: (1) it is purely legal; (2) it does not
implicate the agency’s expertise; (3) review would be de novo;
and (4) no fact-finding is necessary. In those circumstances,
the agency’s decision would be nothing more than an “idle and
useless formality,” rendering remand futile and unnecessary.
NLRB v. Wyman-Gordon Co., 394 U.S. 759, 766 n.6 (1969).

Here, each of those criteria is satisfied. First, questions
of divisibility and application of the categorical approach are
purely legal questions. Singh, 839 F.3d at 282; see also Javier
v. Att’y Gen., 826 F.3d 127, 130 (3d Cir. 2016); Jean-Louis,
582 F.3d at 466. Second, these questions require no special

8
agency expertise. As in Jean-Louis, the BIA has already
exercised its expertise by identifying the mens rea for the
“crime of stalking” in the INA, see Sanchez-Lopez II, 27
I. & N. Dec. at 258, and the application of the categorical
approach itself is not a question “that statutes place primarily
in agency hands,” Jean-Louis, 582 F.3d at 469 (citation
omitted); see Singh, 839 F.3d at 282. The only other issue, the
divisibility of Section 2709.1(a)(1), is one of state law, and thus
falls within our expertise, not the agency’s. See Salmoran, 909
F.3d at 77 (“‘[W]e owe no deference to the BIA’s
interpretation of a state criminal statute,’ which does not entail
the BIA’s special expertise[.]” (quoting Javier, 826 F.3d at
130)). Third, precisely because these are purely legal questions
that do not implicate the agency’s expertise, we would review
them de novo. Salmoran, 909 F.3d at 76; Singh, 839 F.3d at
282. And finally, resolving these questions does not require
the evaluation of any additional evidence because “all relevant
evidence regarding the conviction ha[s] been presented to the
BIA in earlier proceedings.” Fregozo v. Holder, 576 F.3d
1030, 1039 (9th Cir. 2009); see also Ventura, 537 U.S. at 18.

Just as in Singh and Jean-Louis, remand under these
circumstances “would be of no purpose and would lead to an
unnecessary expenditure of time and resources,” Mandujano-
Real, 526 F.3d at 589, so we proceed to the merits of
Vurimindi’s petitions.

B. Whether the Statute Qualifies as a Removable
Offense

To their credit, the parties here both recognize and
acknowledge that Pennsylvania’s stalking statute is a
removable offense if it is divisible, and it is not if it is

9
indivisible.2 For those who may be unfamiliar with
significance of the statute’s divisibility for our analysis,
however, we briefly explain the modification to the categorical
approach that the Supreme Court has authorized in the case of
divisible statutes before turning to the dispositive issue in this
case: whether Pennsylvania’s statute is, in fact, divisible.

The Supreme Court has provided two approaches to
analyzing whether a state conviction qualifies as removable
under the INA. If the state offense is defined by a single set of
elements, we apply the categorical approach, even where the
statute provides different means by which one or more of those
elements may be satisfied. See Rosa v. Att’y Gen., 950 F.3d
67, 75 (3d Cir. 2020). Under the categorical approach, we
compare the conduct covered by the elements of
Pennsylvania’s stalking statute to conduct covered by the
elements of a crime of stalking under the INA to determine “if
[the] state statute’s elements define a crime identical to or
narrower than the generic crime.” Larios v. Att’y Gen., 978
F.3d 62, 67 (3d Cir. 2020) (citing Descamps, 570 U.S. at 261).

2
Vurimindi concedes that if the Pennsylvania statute is
divisible, the offense with the alternative mens rea of “intent
to place [the victim] in reasonable fear of bodily injury” would
be a removable offense. 18 Pa. Stat. and Cons. Stat.
§ 2709.1(a)(1). Even then, however, Vurimindi claims his
conviction was not for a removable offense because, under the
modified categorical approach, the record would not show he
was found guilty of stalking with that intent, as opposed to the
non-removable alternative of intent to place the victim in fear
of nonphysical injury. See Opening Br. 36–37. As we
conclude the statute is indivisible, we need not address this
argument.

10
We “focus[] on the underlying criminal statute rather than the
alien’s specific act . . . to ascertain the least culpable conduct
necessary to sustain a conviction under the statute.” Javier,
826 F.3d at 130 (citations omitted). So if the state statute
“covers more conduct” than the generic offense in the INA,
then it is not a categorical match because it is “overbroad and
does not match the generic offense.” Larios, 978 F.3d at 67.

If, on the other hand, the statute “list[s] elements in the
alternative,” so that it “define[s] multiple crimes,” Mathis v.
United States, 579 U.S. 500, 505 (2016), then the statute is
“divisible” and we apply the modified categorical approach,
Hillocks v. Att’y Gen., 934 F.3d 332, 339 (3d Cir. 2019)
(citation omitted).3 Under that approach, we first “determine
which of the alternative elements was the actual basis for the
underlying conviction,” Evanson v. Att’y Gen., 550 F.3d 284,
291 (3d Cir. 2008),4 and we then “compare that crime, as the

3
As the Supreme Court has explained, elements are “the
‘constituent parts’ of a crime’s legal definition—the things the
‘prosecution must prove to sustain a conviction.’” Mathis, 579
U.S. at 504 (quoting Elements of Crime, Black’s Law
Dictionary 634 (10th ed. 2014)).
4
Under the modified categorical approach, we consider
“extra-statutory materials” to determine whether the
conviction, for example, “was [for] the version of the crime in
the [state] statute . . . corresponding to” a removable crime
under the INA. Descamps, 570 U.S. at 262–63. Appropriate
materials can include the “charging document, written plea
agreement, transcript of plea colloquy, and any explicit factual
finding by the trial judge to which the defendant assented,” but

11
categorical approach commands, with the relevant generic
offense,” Mathis, 579 U.S. at 506.

The generic offense here, of course, is the “crime of
stalking” under 8 U.S.C. § 1227(a)(2)(E)(i). Although the INA
itself does not define the elements of that generic offense, see
id., the BIA has done so, interpreting the term “crime of
stalking,” to require proof of three elements: (1) that the
defendant engaged in conduct more than once, (2) that he
directed the conduct at a specific victim, and, importantly for
our purposes, (3) that he acted with intent to cause the victim
“fear of bodily injury or death.” Matter of Sanchez-Lopez
(“Sanchez-Lopez I”), 26 I. & N. Dec. 71, 74 (BIA 2012)
(emphasis added). And because the INA is a statute that the
BIA is “charged with administering,” and the term “stalking”
is not plain and unambiguous, we accord “deference [to] the
BIA’s reasonable interpretation[].” De Leon-Ochoa v. Att’y
Gen., 622 F.3d 341, 348 (3d Cir. 2010).

Under that interpretation, however, a state statute that
criminalized stalking with the intent to cause “fear [of]
nonphysical injury” would be overbroad and would not support
removal under the INA. Sanchez-Lopez II, 27 I. & N. Dec. at
260–61 (emphasis added); see also In re Shaban, 2018 WL
3045823, at *2 (BIA May 1, 2018) (finding no categorical
match with a state stalking statute criminalizing conduct other
than with the intent to cause fear of physical injury). That
brings us to the dilemma confronting the Government in this
case.

not “police reports or complaint applications.” Shepard v.
United States, 544 U.S. 13, 16 (2005).

12
 By its terms, the Pennsylvania’s statute matches the first
two elements—the actus reus—of the federal generic offense,
but as for mens rea, it criminalizes intent to cause nonphysical,
as well as physical injury because it covers “circumstances
which demonstrate either an intent to place [the victim] in
reasonable fear of bodily injury or to cause substantial
emotional distress to such [a] person.” 18 Pa. Stat. and Cons.
Stat. § 2709.1(a)(1) (emphasis added). Thus, if “reasonable
fear of bodily injury” and “substantial emotional distress” are
simply two means to satisfy a single mens rea element, the
categorical approach applies and the Pennsylvania statute is
not a categorical match because Section 2709.1(a)(1) would
then “sweep[] more broadly” than stalking under the INA.
United States v. Scott, 14 F.4th 190, 197 (3d Cir. 2021); see
also Larios, 978 F.3d at 67. If Section 2709.1(a)(1) is divisible,
on the other hand, then the modified categorical approach
applies and the first alternative mens rea—intent “to place .
. . in reasonable fear of bodily injury”—is a categorical match.
See Sanchez-Lopez II, 27 I. & N. Dec. at 258. For us, then, to
divide or not to divide—that is the question.

C. Whether Pennsylvania’s Statute Is Divisible

In determining whether Pennsylvania’s stalking statute
is divisible, our first task is “to determine whether its listed
items are elements or means.” Singh, 839 F.3d at 283 (quoting
Mathis, 579 U.S. at 517). To distinguish between elements and
means, we determine whether “the [text of the] statute is clear
on its face” and examine whether there are “prior state court
decisions definitively answering the question.” Rosa, 950 F.3d
at 81 (citation omitted). If the statutory text and state court
decisions are insufficiently clear, we “may look ‘to a limited
class of documents (for example, the indictment, jury
instructions, or plea agreement and colloquy).’” Id. (quoting

13
Mathis, 579 U.S. at 505). But if those documents also do not
“speak plainly,” then “the record will not be able to satisfy the
‘demand for certainty’ required” to demonstrate that the
“defendant was convicted of [the] generic offense.” Id. at 82
(quoting Mathis, 579 U.S. at 519). We consider these
authorities in order of priority.

1. Text and Structure of Section
2709.1(a)(1)

We start with the text of Section 2709.1(a)(1), which
provides in pertinent part that a person engages in the crime of
stalking when they:

engage[] in a course of conduct or
repeatedly commit[] acts toward
another person, . . . under
circumstances which demonstrate
either an intent to place such other
person in reasonable fear of bodily
injury or to cause substantial
emotional distress to such other
person[.]

18 Pa. Stat. and Cons. Stat. § 2709.1(a)(1). Parsing the text, a
person only commits the crime when they act—i.e., “engage[]
in a course of conduct or repeatedly commit[] acts” that are
directed “toward another person”5—in a certain manner—i.e.,

5
While Section 2709.1(a)(1) is also worded
disjunctively as to actus reus, neither party argues that it is
divisible as to this element and we need not consider whether
“course of conduct” and “repeatedly commits acts” are

14
“under circumstances which demonstrate” the statutorily
enumerated intent. Id. And § 2709.1(a)(1) is worded
disjunctively with respect to that intent, deploying the
correlative conjunctives “either . . . or” to link the
grammatically parallel infinitive phrases (1) “to place such
other person in reasonable fear of bodily injury,” and (2) “to
cause substantial emotional distress to such other person.” Id.

It is this disjunctive wording as to intent to which the
Government points as textual evidence of divisibility. See
generally Answering Br.; see also Oral Arg. Tr. 25:57–26:36.
But that alone is not conclusive, for while disjunctives in
statutes can provide “textual clue[s]” of divisibility, Hillocks,
934 F.3d at 343, they are not dispositive because statutes that
merely “enumerate[] various factual means of committing a
single element” are not in fact divisible, Mathis, 579 U.S. at
506. And here, other textual clues indicate that § 2709.1(a)(1)
is in fact indivisible and that the enumerated intents are
alternative means, not elements.

First, the statute does not repeat the phrase “an intent.”
This suggests a single intent element that can be
“demonstrate[d]” through one of two means, 18 Pa. Stat. and
Cons. Stat. § 2709.1(a)(1), rather than two, alternative intent
elements, as words are to be given the meaning that proper
grammar assigns them, see Antonin Scalia & Bryan A. Garner,
Reading Law: The Interpretation of Legal Texts § 17 (2012).
If it did repeat “intent,” such that it read “either an intent to
place . . . or an intent to cause . . . ,” then the verb
“demonstrate” would take two direct objects, stated in the

alternative means of satisfying an actus reus element, or two
alternative means of satisfying a single element.

15
disjunctive—the noun phrases “an intent to place [in fear of
bodily harm]” and “an intent to cause [emotional distress],”
which would support the Government’s argument that the two
intents are elements rather than means. But the statute has a
single direct object—“an intent”—with disjunctive
modifiers—“to place [in fear of bodily harm]” and “to cause
[emotional distress]”—that are grammatically parallel, 18 Pa.
Stat. and Cons. Stat. § 2709.1(a)(1), which supports
Vurimindi’s argument that the modifiers are simply different
means of satisfying a single intent element.

Second, the legislature’s choice of the verb
“demonstrate” in Section 2709.1(a)(1) is another strong textual
signal that the alternative intents are means, not elements. A
person must engage in the prohibited conduct “under
circumstances which demonstrate” either of the alternative
intents in order to violate the statute. 18 Pa. Stat. and Cons.
Stat. § 2709.1(a)(1). And the plain meaning of “demonstrate”
at the time of enactment was “to show clearly” or “to illustrate
or explain esp[ecially] with many examples,” Demonstrate,
Merriam-Webster’s Collegiate Dictionary 308 (10th ed. 1993);
see also Demonstrate, Oxford English Dictionary 447 (2d ed.
1989) (“[t]o point out, indicate; to exhibit, set forth”).6 Thus,
by describing the two alternatives as means of
“demonstrate[ing]” intent, the statute communicates that they
are illustrative, explanatory, and exhibitive of a separate and

6
The word “demonstrate” was first included in
Pennsylvania’s stalking statute in 1993. See 18 Pa. Stat. and
Cons. Stat. § 2709 (1993); 1993 Pa. Legis. Serv. 124 (West);
see also B. Benjamin Haas, The Formation and Viability of
Anti-Stalking Laws, 39 Vill. L. Rev. 1387, 1401 (1994) (noting
that Pennsylvania’s anti-stalking statute was enacted in 1993).

16
singular subject. They are, in short, means of proving the
requisite mens rea, not elements.

Additional support for a finding of indivisibility comes
from our decision in United States v. Gonzalez, 905 F.3d 165
(3d Cir. 2018), interpreting the federal cyberstalking statute, 18
U.S.C. § 2261A(2). In Gonzalez, the defendants argued that
the district court erred in failing to give a jury instruction
requiring, inter alia, that the jury unanimously agree on the
defendants’ mens rea to sustain a conviction under that statute.
Like the Pennsylvania stalking statute, the federal
cyberstalking statute utilizes the disjunctive to describe a
singular “intent,” requiring the Government to establish that
the defendant:

engaged in a course of conduct that
placed a person in reasonable fear
of death or serious bodily injury, or
causes substantial emotional
distress . . . “with the intent to kill,
injure, harass, intimidate, or place
under surveillance with intent to
kill, injure, harass, or intimidate”
that person.

905 F.3d at 183 (quoting 18 U.S.C. § 2261A(2)).

After reviewing the text of the statute and caselaw
interpreting it, we could discern no evidence that Congress
“intended to create separate offenses for stalking ‘with the
intent to kill’ as opposed to stalking ‘with the intent to . . .
injure’ or ‘with the intent to . . . harass,’” and therefore held
that the different mental states in the statute constitute alternate
factual means and not alternate elements. Id. at 185. In

17
reaching this conclusion, we found especially persuasive the
fact that the cyberstalking statute “sets forth different tiers of
punishment based not on the mental state of the defendant, but
on the harm suffered by the victim,” reflecting that their mental
state must be evidenced by “equivalent blameworthiness or
culpability.” Id. at 185–86 (quoting United States v. Yeaman,
194 F.3d 442, 454 n.6 (3d Cir. 1999)).

Here, the Pennsylvania legislature drafted its stalking
statute in a grammatically similar way, requiring a singular
“intent” with disjunctive modifiers, and it also did not make
punishment contingent on the intent of the defendant. Rather,
the grade of the offense and the punishment for a violation of
§ 2709.1 are dependent on an offender’s prior criminal history.
See 18 Pa. Stat. and Cons. Stat. § 2709.1(c).7 So, just as in
Gonzalez, as long as the defendant’s conduct “was taken with
an intent to cause the victim harm, the specific mental state
does not make a difference to the defendant’s culpability.”
Gonzalez, 905 F.3d at 185. Thus, here too, that the sentencing
scheme indicates that the different intent one “means” with
“equivalent blameworthiness,” id. (quoting Yeaman, 194 F.3d
at 454 n.6), along with the text and grammar, favor a finding
of different means, not different elements.

7
Under the Pennsylvania stalking statute, an offense
can be a first-degree misdemeanor or third-degree felony,
depending on offender’s criminal history. See 18 Pa. Stat. and
Cons. Stat. § 2709.1(c).

18
 2. Pennsylvania Courts’ Interpretation of
Section 2709.1(a)(1)

While parties may point to state court decisions as
“authoritative sources of state law” that resolve a means-or-
elements question, Mathis, 579 U.S. at 518, the Government
has not identified any case that squarely addresses this question
in relation to Section 2709.1(a)(1), Answering Br. 24–25, nor
have we.8 If anything, the smattering of cases discussing the
elements of a Section 2709.1(a)(1) conviction suggest instead
that Vurimindi has the stronger argument.

For instance, the Government points us to the Superior
Court’s statement in Commonwealth v. Abed that the “intent to
place one in fear of bodily injury is but one mens rea that will
sustain a conviction under § 2709[] . . . as a conviction may be
upheld upon a showing that the accused intended to harass,
annoy, alarm[,] or cause substantial emotional distress.” 989
A.2d 23, 27 (Pa. Sup. Ct. 2010) (citation omitted). Answering
Br. 24–25. But Abed is itself quoting the Superior Court’s
decision in Commonwealth v. Miller, 689 A.2d 238, 240 (Pa.
Super. Ct. 1997), and Miller actually undermines the

8
To determine whether an “authoritative source[] of
state law,” Mathis, 579 U.S. at 518, has resolved “this means-
or-elements question,” we try to determine how the
Pennsylvania Supreme Court would decide the substantive
legal issue, Singh, 839 F.3d at 283 & n.5. However, when, as
here, “there is no opinion or other ‘persuasive data’ on point
from the Supreme Court of Pennsylvania,” we may rely on
authoritative decisions of the Superior Court of Pennsylvania.
Id. at 283 n.5 (quoting Sheridan v. NGK Metals Corp., 609
F.3d 239, 254 (3d Cir. 2010)).

19
Government’s reading, describing a singular “intent element,”
with alternative means of establishing that element. See Miller,
689 A.2d at 240 n.1 (“The intent element of stalking requires
proof that the actor either intended to place the victim in
reasonable fear of bodily injury or to cause substantial
emotional distress.” (emphasis added)). In any event, when
read in context, Abed and Miller, which both involve
sufficiency of the evidence challenges on direct appeal, suggest
only that a Section 2709.1(a)(1) stalking conviction can be
upheld where the Government proves that the defendant acted
under circumstances demonstrating one of the two enumerated
intents—not that each of the intents listed is a separate element
of two alternative stalking crimes. See Miller, 689 A.2d at
239–41; Abed, 989 A.2d at 27–28.9

9
The Government also cites two non-precedential
decisions: Commonwealth v. Heflin, No. 1700 WDA 2013,
2014 WL 10588385, at *1 (Pa. Super. Ct. Sept. 30, 2014) (non-
precedential), and Commonwealth v. Hoffman, No. 2277 EDA
2015, 2016 WL 7176962, at *1 (Pa. Super. Ct. Dec. 9, 2016)
(non-precedential), as evidence of a divisible intent element.
Answering Br. 25. Under Pennsylvania Court Rules, Pa. I.O.P.
65.37, we may consider such unpublished state court opinions
as persuasive authority “when predicting state law,” Taransky
v. Sec’y of U.S. Dep’t of Health & Hum. Servs., 760 F.3d 307,
317 n.9 (3d Cir. 2014), but neither case is persuasive on this
point. In Heflin, the Superior Court drew a distinction between
the lower intent required for the crime of harassment and the
higher “malevolent intent” required for stalking. 2014 WL
10588385, at *1. Nothing in that decision suggests that the
intent element of the stalking statute is divisible; rather, the
court simply addressed the sufficiency of the evidence issue
raised by the defendant on appeal. Id. And in Hoffman—which

20
 Similarly, although Pennsylvania courts have not
squarely considered the divisibility of Section 2709.1(a)(1),
those discussing the elements of a conviction uniformly
describe the mens rea element as a singular “intent.” See, e.g.,
Commonwealth v. Danzey, 210 A.3d 333, 342 (Pa. Super. Ct.
2019) (“[B]oth Harassment and Stalking have an intent
element.” (emphasis added)), appeal denied, 219 A.3d 597 (Pa.
2019); Commonwealth v. D’Collanfield, 805 A.2d 1244, 1248–
49 (Pa. Super. Ct. 2002) (interpreting a “virtually identical”

involved the admission of a 2003 stalking conviction as
character evidence at trial—the trial court referred to “either
section of [the] stalking [statute],” which the Government takes
to establish two alternative intent elements. Answering Br. 25
(citing Hoffman, 2016 WL 7176962, at *3). Effective in mid-
2003, the Pennsylvania legislature repealed and replaced the
predecessor stalking statute with Section 2709.1. The previous
statute enumerated the alternative intents in two numbered sub-
sections. 18 Pa. Cons. Stat. Ann. § 2709(b)(1)–(2) (1999). The
2003 conviction at issue in Hoffman dealt with the previous
version of the stalking statute, see Appellant’s Br.,
Commonwealth v. Hoffman, 2016 WL 5944544, at *18 (Pa.
Super. Ct. July 26, 2016), and thus the court’s reference to
“either section” of the statute literally refers to two subsections
of the now-repealed stalking statute. In any event, the
defendant in Hoffman argued that her prior conviction for
stalking could not be probative of violence absent evidence of
the underlying facts of her conviction, and the Superior Court
flatly rejected this argument. 2016 WL 7176962 at *4. Thus,
Hoffman does not imply—much less persuasively reason—that
the alternative intents in Pennsylvania’s stalking statute are
elements rather than means.

21
stalking statute and holding that “stalking requires a repetitive
course of malevolent conduct, the intent of which was to place
someone in fear of bodily injury or cause substantial emotional
distress” (emphasis added)); see also Commonwealth v.
Rushing, 236 A.3d 1098 (Pa. Super. Ct. 2020) (non-
precedential table decision) (“[S]talking requires an intent to
create fear or emotional distress in the victim, an element not
required to establish contempt.” (emphasis added)), appeal
denied, 239 A.3d 2 (Pa. 2020); Commonwealth v. Mongeau,
240 A.3d 153 (Pa. Super. Ct. 2020) (non-precedential table
decision) (describing the intent element as the intent as “to
place [the victim] in fear of bodily injury and/or to cause her
substantial emotional distress.” (emphasis added)), appeal
denied, 244 A.3d 1225 (Pa. 2021).

3. Additional Evidence of Indivisibility

As a last resort, we “may look ‘to a limited class of
documents (for example, the indictment, jury instructions, or
plea agreement and colloquy),’” Rosa, 950 F.3d at 81 (quoting
Mathis, 579 U.S. at 505), and here, those documents also
support the conclusion that the statute is indivisible—even
when the Government contends otherwise. See Answering Br.
26–30.

The documents specific to Vurimindi’s conviction—
including the criminal information, order of sentence, and trial
disposition form—are a good illustration. The Government
asserts that their references to “Stalking—Repeatedly
Commits Acts to Cause Fear,” as shorthand for Vurimindi’s
Section 2709.1(a)(1) conviction prove that Pennsylvania
courts treat the two intents as giving rise to two different
offenses. Answering Br. 28; see also JA131, 133–34. But its
hypothesis, which is that the “cause fear” shorthand refers only

22
to stalking with intent to “cause fear of bodily injury,”
relegating the distinct offense of stalking with intent to cause
emotional distress to a different shorthand, Answering Br. 27–
29, is not borne out by a survey of the case law or dockets in
Pennsylvania stalking cases.

To the contrary, that survey reveals that the same “cause
fear” shorthand that appears in Vurimindi’s paperwork is used
in all Section 2709.1(a)(1) stalking cases, including those
involving intent to cause substantial emotional distress. As just
one example, though the Superior Court’s opinion in
Commonwealth v. Sammy makes clear that the defendant’s
conduct involved only “intent to cause substantial emotional
distress,” No. 1671 WDA 2016, 2018 WL 2010436, at *5 (Pa.
Super. Ct. Apr. 30, 2018) (non-precedential), the docket
repeatedly uses the shorthand “Stalking—Repeatedly Commit
Acts To Cause Fear.” See Commonwealth v. Sammy, No. CP-
02-CR-0004751-2015 (Ct. Common Pleas, Allegheny Cnty.
Oct. 6, 2016), available at
https://ujsportal.pacourts.us/Report/CpDocketSheet?docketN
umber=CP-02-CR-0004751-
2015&dnh=WxFwQeZfNC5vOfV1S0%2Bu8A%3D%3D;
see also Reply Br. 7 n.2 (collecting other examples).
Pennsylvania’s sentencing guidelines likewise use “Stalking—
repeated acts to cause fear of BI [bodily injury]” as shorthand
for all Section 2709.1(a)(1) cases, regardless of intent. See 204
Pa. Code § 303.15.

The Government also posits that the model jury
instructions for Section 2709.1 support a finding that the
statute is divisible as to intent.10 Answering Br. 26. We draw

10
We have previously considered Pennsylvania’s
pattern jury instructions to be relevant to our divisibility

23
the opposite conclusion. The relevant jury instructions are
structured in a way that communicates only two elements—
one actus reas and one mens rea:

To find the defendant guilty of this offense, you
must find that each of the following elements has
been proven beyond a reasonable doubt:

First, that the defendant:

a. engaged in a course of conduct . . . or

b. repeatedly committed acts . . . or

c. repeatedly communicated . . . and

Second, that the defendant did so under
circumstances that demonstrated that [he] [she]
intended:

a. to put [name of victim] in reasonable
fear of bodily injury; or

analysis, see, e.g., United States v. Steiner, 847 F.3d 103, 119
(3d Cir. 2017); however, the Pennsylvania Supreme Court has
explained that “[t]he Suggested Standard Jury Instructions
themselves are not binding . . . ; rather, as their title suggests,
the instructions are guides only,” Commonwealth v. Eichinger,
108 A.3d 821, 845 (Pa. 2014). Thus, at most, the pattern jury
instructions are but one relevant source of authority that we
consider in determining whether Section 2709.1(a)(1) is
divisible.

24
 b. intended to cause [him] [her]
substantial emotional distress.

Pa. S.S.J.I. (Crim), § 15.2709.1. Even aside from enumerating
only two elements, the “[f]irst,” the act, and the “[s]econd,” the
intent, id., a subcommittee note to these jury instructions
explains that “[i]n a prior version of these instructions, the
subcommittee had suggested that a third element of this
offense required that the Commonwealth prove that the victim
was actually placed in fear by the defendant’s actions,” and that
because the proposed third element appeared neither in the
statute nor in the decisions of Pennsylvania’s appellate courts,
the subcommittee removed it. Id. (emphasis added). Thus,
directly contrary to the Government’s reading, the pattern jury
instructions make explicit that the alternative intents are simply
means of satisfying the intent element.

In sum, the Government has identified no evidence that
supports divisibility. The statute, the case law, and the
available state court documents all support the opposite
conclusion.11 Because Pennsylvania’s stalking statute is
indivisible as to intent, we apply the categorical approach. And
under the categorical approach, Section 2709.1(a)(1), which
sweeps more broadly than its generic counterpart in the INA,
is not a categorical match. Vurimindi’s offense of conviction
therefore does not qualify as a removable offense.

11
Because there was no indictment, plea, or jury trial in
Vurimindi’s underlying criminal case, there are no additional
documents associated with those events to review.

25
IV. Conclusion

For the foregoing reasons, we will grant Vurimindi’s
petitions, vacate the BIA’s orders, and remand to the agency
for further proceedings consistent with this opinion.

26