Roberson v. Texas

_________________
_________________
1 Cite as: 604 U. S. ____ (2024)
Statement of S
OTOMAYOR, J.
SUPREME COURT OF THE UNITED STATES
No. 24–5753 (24A349)
ROBERT LESLIE ROBERSON III v. TEXAS
ON APPLICATION FOR STAY AND ON PETITION FOR A WRIT OF
CERTIORARI TO THE TEXAS COURT OF CRIMINAL APPEALS
[October 17, 2024]
The application for stay of execution of sentence of death
presented to J
USTICE ALITO and by him referred to the
Court is denied. The petition for a writ of certiorari is de-
nied.
Statement of J
USTICE SOTOMAYOR respecting the denial
of the application for stay of execution and denial of certio-
rari.
In 2003, a Texas jury convicted Robert Leslie Roberson of
murdering his chronically ill infant daughter, Nikki, and
sentenced him to death. At trial, the State relied heavily
on expert testimony to show that “child physical abuse con-
sistent with the picture of what they call shaken impact
syndrome,” or shaken baby syndrome, had caused Nikki’s
death. Tr. 55 (Feb. 3, 2003). At the time, infants exhibiting
a triad of symptoms—subdural bleeding, brain swelling,
and retinal hemorrhages—were presumed to be victims of
child abuse, absent evidence of a major traumatic event
such as a car crash. This view went so unquestioned that
Roberson’s own defense counsel told the jury in his opening
statement that this was “unfortunately a shaken baby
case,” that the evidence would “show that Nikki did suffer
injuries that are totally consistent with those applied by ro-
tational forces more commonly known as shaken baby syn-
drome,” and that he would not “tell you that there is just no
responsibility here at all.” Id., at 57–58.
As the Texas Court of Criminal Appeals (TCCA) has itself

2 ROBERSON v. TEXAS
Statement of S
OTOMAYOR, J.
confirmed, the scientific basis for shaken baby syndrome
has since been called into significant question. Were the
experts at Roberson’s trial to testify to the same theory
again today, they “would be confronted with twenty years
of reputable scientific evidence that contradicts their trial
testimony.” Ex parte Andrew Wayne Roark, No. WR–
56,380–03, 2024 WL 4446858, at *26 (Tex. Ct. Crim. App.,
Oct. 9, 2024). For that reason, the TCCA just this week
granted a new trial to Andrew Wayne Roark, a non-capital
defendant whose child-abuse conviction rested on the same
shaken-baby-syndrome testimony, from the same expert
witness, that led to Roberson’s conviction. When Roberson
sought a stay of execution based on the argument the TCCA
credited in Roark, the TCCA summarily denied relief.
Roberson now seeks a stay of execution from this Court,
but he presents no cognizable federal claim. Meanwhile,
mounting evidence suggests that Nikki died from a combi-
nation of pneumonia and improperly prescribed medica-
tion, that Nikki’s severe bruising could well have occurred,
at least in part, as a result of her extensive medical proce-
dures at the hospital, and that Roberson (who repeatedly
sought emergency medical care for his daughter in the days
and hours leading up to her death) committed no crime at
all. Because this Court is powerless to act without a color-
able federal claim, and because the Texas Board of Pardons
and Paroles declined to recommend clemency, only one rem-
edy remains: an executive grant of a reprieve delaying Rob-
erson’s execution by thirty days.
I
A
In January 2002, Roberson discovered that he was the fa-
ther of a young child, who had then just turned two and
resided with her maternal grandparents. Roberson took
custody of Nikki and cared for her together with his par-

3 Cite as: 604 U. S. ____ (2024)
Statement of S
OTOMAYOR, J.
ents. Nikki suffered from several serious illnesses, includ-
ing chronic antibiotic-resistant infections and unexplained
“breathing apnea.” Pet. App. 208a–210a, 285a, 312a. On
January 28, 2002, Roberson and his mother took Nikki to
the hospital following five days of vomiting, coughing, and
diarrhea; an attending physician prescribed Phenergan, a
potent anti-nausea medication, and sent Nikki back home.
As Nikki’s fever continued to rise, Roberson brought her
back to the hospital the next morning. Although Nikki now
had a 104.5 degree fever, the attending physicians pre-
scribed Phenergan with Codeine (a cough medication), and
again sent Nikki home. The FDA has since warned against
prescribing these medications to young children. Phener-
gan, the FDA warns, “should not be used in pediatric pa-
tients less than 2 years of age because of the potential for
fatal respiratory depression.” Codeine, too, “carr[ies] seri-
ous risks, including slowed or difficult breathing and death,
which appear to be a greater risk in children younger than
12 years.”*
When Roberson woke up the next morning, he found
Nikki turning blue and unable to breathe. After attempting
to revive her, Roberson called emergency services. He told
police that Nikki had fallen out of bed earlier that night,
that he had comforted her and sat with her for two hours
before falling back asleep, and that he had found her unre-
sponsive when he next woke up. Tr. 170 (Feb. 3, 2003).
Over the course of two days, doctors made extensive efforts
to treat Nikki, including by surgically affixing a pressure
——————
*FDA, FDA-Approved Drugs, Phenergan, online at https://www.
accessdata.fda.gov/drugsatfda_docs/label/ 2004/07935s030lbl.pdf (last
accessed Oct. 17, 2024); FDA, FDA Drug Safety Communication: FDA
requires labeling changes for prescription opioid cough and cold medi-
cines to limit their use to adults 18 years and older (Apr. 20, 2017),
https://www.fda.gov/ drugs/drug-safety-and-availability/fda-drug-safety-
communication-fda-restricts - use -prescription-codeine-pain-and-cough-
medicines-and (last accessed Oct. 17, 2024).

4 ROBERSON v. TEXAS
Statement of S
OTOMAYOR, J.
monitor to the top of her skull. Tragically, these efforts
could not save Nikki’s life.
Medical personnel diagnosed Nikki with shaken baby
syndrome based on the “triad” of symptoms: subdural bleed-
ing, brain swelling, and retinal hemorrhages. An autopsy
revealed evidence of internal bleeding near Nikki’s brain,
as well as extensive bruising on her head, face, and shoul-
ders, leading the doctor who conducted the autopsy to con-
clude that Nikki had died of “blunt force head injuries.” Tr.
55 (Feb. 3, 2003).
Texas charged Roberson with capital murder. At trial,
the State repeatedly argued that “Nikki died or . . . was the
victim of child physical abuse consistent with the picture of
what they call shaken impact syndrome” or shaken baby
syndrome. Id., at 54–55. Notwithstanding Roberson’s in-
sistence that he was innocent and the abundant evidence
that Nikki had been severely ill prior to her death, Rob-
erson’s trial counsel conceded that his client had abused
Nikki and argued only that Roberson had not intended to
kill her. Id., at 60–61.
The State called Dr. Janet Squires as an expert witness
to substantiate its theory that Nikki had died of shaken
baby syndrome. Dr. Squires testified that Nikki’s extensive
brain injuries could not have been “explained by a simple
impact.” Tr. 107 (Feb. 4, 2003). Instead, she said, it was
her opinion “that there was some component of shaking
that happened to explain all the deep brain injury out of
proportion, I would say, to the injury to the skull and the
back of the head.” Id., at 107–108. Dr. Squires further tes-
tified that children can “stop breathing immediately” if “you
shake them hard enough and you hurt them bad enough.”
Id., at 108. Both Dr. Squires and Dr. Urban (the physician
who had conducted the autopsy) also testified to the bruis-
ing on Nikki’s body. Because they examined Nikki only af-
ter her treatment at the hospital, however, their testimony
left unclear precisely when Nikki sustained these injuries.

5 Cite as: 604 U. S. ____ (2024)
Statement of S
OTOMAYOR, J.
See e.g., Tr. 72 (Feb. 5, 2003) (Dr. Urban acknowledging
medical treatment could have caused some of the bruising);
id., at 96 (similar); id., at 79 (Dr. Urban noting that she was
“not prepared to say” bruises on Nikki’s arm and foot had
been caused by Roberson). In closing, the prosecution relied
on this expert testimony to argue again that Nikki died of
“shaken baby syndrome” and “shaken impact syndrome.”
Tr. 25 (Feb. 11, 2003). The jury convicted Roberson and
sentenced him to death.
B
In 2016, Roberson filed a habeas petition under a Texas
statute that allows a court to grant habeas relief when new
scientific evidence casts doubt on a conviction. Tex. Code.
Crim. Proc. art. 11.073. In his petition, Roberson presented
extensive evidence, including in the form of expert testi-
mony, that the medical community no longer recognized the
“triad” as presumptive evidence of child abuse. Expert
opinion, rather, supported the conclusion that Nikki’s only
external injuries were consistent with a short fall from her
bed, and that the extensive subdural bleeding found at the
autopsy (which led to the conclusion that Nikki had died
from blunt force impacts to her head) had in fact been a re-
sult of her surgical treatment in the hospital. Further sup-
porting that theory, Roberson presented CAT scans taken
of Nikki’s head soon after she initially arrived at the hospi-
tal, which showed only a very small amount of subdural
bleeding.
At an evidentiary hearing, multiple specialists further
testified that Nikki’s lungs showed severe pneumonia and
that she had a toxic level of respiratory-suppressing pre-
scription drugs in her system at the time of her death. Rob-
erson also pointed to the FDA’s urgent warnings against
the use of Phenergan and Codeine in young children “be-
cause of the potential for fatal respiratory depression”
(Phenergan) and the serious risk of “slowed or difficult

6 ROBERSON v. TEXAS
Statement of S
OTOMAYOR, J.
breathing and death” (Codeine). In an order that adopted
large portions of the State’s proposed findings of fact verba-
tim, the trial court denied Roberson relief. It found that no
“false evidence was presented in regards to” shaken-baby
syndrome, and that in any event “no evidence was pre-
sented to the jury that shaking alone” had caused Nikki’s
death. Pet. App. 177a, 181a. The trial court’s findings did
not address the other evidence presented at the evidentiary
hearings. The TCCA summarily affirmed. App. 4a.
In the years that followed, Roberson retained additional
experts who each concluded that Nikki had died of pneumo-
nia and toxic levels of anti-respiratory drugs. Based on that
additional evidence, Roberson filed another state habeas
petition, which the TCCA declined to consider on proce-
dural grounds. This Court denied certiorari. See 144 S. Ct.
129 (mem.) (Oct. 2, 2023).
C
As Roberson sought to prove his innocence, the parallel
case of Andrew Wayne Roark made its way through the
Texas courts. Roark’s case involved claims nearly identical
to Roberson’s, raised under the same provisions of Texas
law. There, as here, the State had argued at trial that
Roark “caused the infant victim serious bodily injury by vi-
olently shaking her and possibly striking her with or
against something, respectively referred to as Shaken Baby
Syndrome and Shaken Impact Syndrome.” Ex parte Roark,
2024 WL 4446858, at *2. There, as here, the State had re-
lied on expert testimony from Dr. Janet Squires about
shaken baby syndrome to secure a conviction. Id., at *3.
There, as here, the defendant challenged his conviction on
the ground that the science underlying the shaken baby
syndrome diagnosis had since been discredited.
In a landmark opinion, the TCCA held that, based on cur-
rent medical science, “there would be a marked shift in tes-
timony today concerning the effect of a short-distance fall

7 Cite as: 604 U. S. ____ (2024)
Statement of S
OTOMAYOR, J.
to a child, the effect of shaking a child . . . retinal hemor-
rhaging, and [shaken baby syndrome] in general as applied
to [the victim’s] injuries.” Id., at *23. Had Roark been tried
on the same theory today, the TCCA explained:
The [State’s] experts would be confronted with twenty
years of reputable scientific studies and publications
that, if graphed, continually point away from their
stated positions. If the expert were to experience the
ostrich effect and wish to bury his or her head in the
sand, then that expert would have to bear the brunt of
a grueling cross-examination. One in which they would
be confronted with twenty years of reputable scientific
evidence that contradicts their trial testimony.
Id., at *26. The TCCA thus granted Roark a new trial. Ibid.
While Roark was pending before the TCCA, Roberson
filed a “suggestion for rehearing” with that court seeking
reconsideration of its prior decision on his habeas petition,
based on the proceedings in Roark and on a statement of
support from 86 members of the Texas House of Represent-
atives. Pet. App. 492–546a, 547–550a. In that statement,
the bipartisan House Criminal Justice Caucus urged the
Board of Pardons and Paroles to recommend clemency in
light of the “strong evidence of [Roberson’s] innocence. Id.,
at 548a. The representatives further explained that the
House had unanimously passed Article 11.073 specifically
to allow “challenges to convictions that were based on dis-
proven or incomplete science,” and that they were “dis-
mayed to learn that this law has not been applied as in-
tended and has not been a pathway to relief—or even a new
trial—for people like Mr. Roberson.” Ibid. Brian Wharton,
the lead detective on Roberson’s case, likewise stated that
he believed Roberson to be innocent and that clemency
would be appropriate. Id., at 549a. Notwithstanding these
statements, the Board of Pardons and Paroles declined to
recommend clemency.

8 ROBERSON v. TEXAS
Statement of S
OTOMAYOR, J.
Before issuing its opinion in Roark, the TCCA again de-
nied relief to Roberson without opinion. In a concurrence,
Justice Yeary maintained that Roberson’s case had not de-
pended exclusively on shaken-baby syndrome evidence be-
cause it had also involved evidence of blunt force impacts,
suggesting Roberson could have beaten Nikki to death. Ex
parte Roberson, Nos. WR–63,081–03, 04 (Tex. Crim. App.
Oct. 10, 2024) (Yeary, J., concurring in denial of suggestion
for reconsideration).
After the TCCA issued Roark, Roberson filed a fourth
post-conviction petition in light of the TCCA’s decision, il-
lustrating in detail that the testimony as to shaken-baby
syndrome in Roark had been nearly indistinguishable from
the testimony in his case. Addressing Justice Yeary’s con-
currence, Roberson pointed out that the CAT scans taken
directly after Nikki’s admission to the emergency room
showed only minimal subdural bleeding, suggesting that
the blunt impact evidence on which Justice Yeary relied
could well have been caused by Nikki’s subsequent brain
surgery, rather than by any injuries she suffered at home.
By a 5-4 vote, the TCCA again denied relief.
II
Current postconviction remedies often fail to correct con-
victions “secured by what we now know was faulty science.”
McCrory v. Alabama, 603 U. S. ___, slip. op., at 1 (2024)
(statement of J
USTICE SOTOMAYOR respecting the denial of
certiorari). This case is emblematic of that problem. To its
credit, Texas has “led the way in forensic science reform in
criminal procedure,” including by passing the statute that
allowed Roberson to challenge his conviction based on
changes in forensic science that seriously undermined the
integrity of his criminal trial. Id., at 11. Tragically, that
statute did not help Roberson in this case.
Roberson now seeks a stay of execution with this Court.
Few cases more urgently call for such a remedy than one

9 Cite as: 604 U. S. ____ (2024)
Statement of S
OTOMAYOR, J.
where the accused has made a serious showing of actual in-
nocence, as Roberson has here. Yet this Court can grant a
stay only if Roberson can show a “significant possibility of
success on the merits” of a federal claim. Hill v.
McDonough, 547 U. S. 573, 584 (2006). Roberson did not
bring federal claims before the TCCA. Instead, he relied
exclusively on Texas’s Article 11.073. Before this Court,
Roberson’s only federal challenge is to the TCCA’s practice
of issuing “boilerplate opinions dismissing subsequent ha-
beas petitions for purported failure to satisfy” Texas’s pro-
cedural requirements in habeas cases. Appl. 6. This Court
has held, however, that it has “no power to tell state courts
how they must write their opinions.” Coleman v. Thomp-
son, 501 U. S. 722, 739 (1991).
Nevertheless, it is notable that the TCCA’s decisions in
this case do not address the whole of Roberson’s evidence of
actual innocence. Justice Yeary stated that shaken-baby
syndrome was not dispositive at Roberson’s trial because
the State introduced evidence of bruising and “multiple im-
pacts to Nikki’s head.” Ex parte Roberson, Nos. WR–
63,081–03, 04, at 2 (Yeary, J., concurring in denial of sug-
gestion for reconsideration). Yet as Roberson explained to
the TCCA, the record reveals that shaken-baby syndrome
was the prosecution’s core theory throughout the trial. The
evidence of bruising to Nikki’s head does not change that,
for two reasons. First, no court has yet considered the evi-
dence suggesting that the “areas of dense hemorrhage” dis-
covered during Nikki’s autopsy, which the examiner took to
be signs of blunt force impact, in fact resulted from Nikki’s
subsequent brain surgery. As already explained, several
experts concluded that CAT scans taken shortly after
Nikki’s admission to the hospital showed only minimal sub-
dural bleeding. Second, the TCCA previously affirmed the
habeas court’s factual finding that “Dr. Janet Squires . . .
clearly stated [at trial] that Nikki’s injuries were unlikely to
have been caused by an impact alone,” but rather were

10 ROBERSON v. TEXAS
Statement of S
OTOMAYOR, J.
caused by shaking and a subsequent impact on a surface
such as a mattress. App. 173a (emphases added). Thus,
the Texas courts themselves determined that the State’s ex-
pert had disavowed a beating or impact-only theory of death
at trial.
Under these circumstances, a stay permitting examina-
tion of Roberson’s credible claims of actual innocence is im-
perative; yet this Court is unable to grant it. That means
only one avenue for relief remains open: an executive re-
prieve. In Texas, as in virtually every other State and the
federal government, “[t]he Executive has the power to exer-
cise discretion to grant clemency and affect sentences at
any stage after an individual is convicted[.]” Vandyke v.
State, 538 S. W. 3d 561, 581 (Tex. Ct. Crim. App. 2017).
Preventing the execution of one who is actually innocent by
means of a review “unhampered by legal technicalities” is
a core historical purpose of the executive power to grant
pardons or reprieves. Christen Jensen, The Pardoning
Power in the American States 49 (1922); see also Herrera v.
Collins, 506 U. S. 390, 417 (1993) (“History shows that the
traditional remedy for claims of innocence based on new ev-
idence, discovered too late in the day to file a new trial mo-
tion, has been executive clemency”); Graham v. Texas Bd.
of Pardons and Paroles, 913 S. W. 2d 745, 748 (Tex. Ct.
Crim. App. 1996) (same). An executive reprieve of thirty
days would provide the Texas Board of Pardons and Paroles
with an opportunity to reconsider the evidence of Rob-
erson’s actual innocence. That could prevent a miscarriage
of justice from occurring: executing a man who has raised
credible evidence of actual innocence.