In the

    United States Court of Appeals
                For the Seventh Circuit
                    ____________________
No. 19‐3158
JOHN MYERS,
                                                Petitioner‐Appellee,
                                v.

RON NEAL,
                                             Respondent‐Appellant.
                    ____________________

        Appeal from the United States District Court for the
        Southern District of Indiana, Indianapolis Division.
         No. 1:16‐cv‐2023 — James R. Sweeney, II, Judge.
                    ____________________

     ARGUED MAY 26, 2020 — DECIDED AUGUST 4, 2020
             AMENDED SEPTEMBER 16, 2020
               ____________________

   Before FLAUM, SCUDDER, and ST. EVE, Circuit Judges.
   SCUDDER, Circuit Judge. Indiana University student Jill
Behrman went for a bike ride one morning but never re‐
turned. The police later found her bicycle less than a mile
from the home of John Myers II, on the north side of Bloom‐
ington. Two years later a woman named Wendy Owings came
forward confessing to the murder, but the case was reopened
when a hunter came upon Behrman’s remains far from the
2                                                     No. 19‐3158

location Owings described. A renewed investigation led the
authorities to Myers, who was eventually charged with the
murder. Six years after Behrman’s disappearance, a jury con‐
victed him. Multiple Indiana courts affirmed. Myers then
sought relief in federal court, and the district court granted
his application for a writ of habeas corpus, concluding that
Myers’s counsel performed so deficiently at trial as to under‐
mine confidence in the jury’s guilty verdict. We reverse.
    The district court was right about the performance of My‐
ers’s trial counsel. It was deficient and plainly so in at least
two ways. What leads us to reinstate Myers’s conviction,
though, is the strength of the state’s case against him separate
and apart from those errors. Among the most convincing evi‐
dence were the many self‐incriminating statements that My‐
ers made to many different people, like telling his grand‐
mother that, if the police ever learned what he did, he would
spend the rest of his life in jail. The weight of these statements,
when combined with other evidence, leads us to conclude
that his counsel’s deficient performance did not prejudice
him. The proper outcome is to respect the finality of Myers’s
conviction in the Indiana courts.
                                 I
    A. The Murder and Investigation
   Jill Behrman disappeared during a morning bicycle ride
on May 31, 2000. Local authorities and the Bloomington com‐
munity sprung to action with assistance from volunteer
search groups, neighboring police forces, state authorities,
and eventually the FBI. The police established the timeline of
that morning: Behrman, a skilled cyclist, planned to go for a
ride before starting work at noon at the University’s Student
No. 19‐3158                                                    3

Recreational Sports Center. She logged off her computer at
9:32 a.m. at her parents’ house, which was close to the center
of town. Two people reported seeing Behrman’s bike lying by
the road near farmland northwest of Bloomington at some
point around noon that day. Nobody could locate her,
though.
    Initial leads pointed quite literally in different directions.
Which way Behrman rode her bike that morning was one of
the unsolved questions in the investigation and became a fo‐
cus of the eventual trial. Everyone agreed that she started her
ride from her parents’ house in Bloomington. Whether she
rode north or south was what mattered. Behrman’s riding
north was important to the theory the state would present at
trial because it placed her near the home of John Myers. But
some early leads suggested that Behrman rode south that
morning. The Appendix contains a map with markings of the
locations pertinent to the case.
    Myers lived about a mile from where Behrman’s bike was
found on North Maple Grove. Given this proximity, Bloom‐
ington Detective Rick Crussen interviewed him on June 28,
2000. Myers stated that he had been on vacation the week of
Behrman’s disappearance. He added that he had been “here
and there” but mainly at home because his plans to take a trip
with his girlfriend Carly Goodman had fallen through. While
checking Myers’s explanations, the authorities learned that
his relationship with Goodman, a high school senior at the
time, ended a few weeks earlier than he had described. Good‐
man also told the police that she had no plans to go anywhere
with Myers.
   In 2002, a woman named Wendy Owings came forward
and confessed to Behrman’s murder. Owings, a Bloomington
4                                                  No. 19‐3158

resident, was facing unrelated felony charges when authori‐
ties interviewed her and asked her whether she knew about
the Behrman disappearance—which by then was widely
known around town. Owings faced up to 86 years’ imprison‐
ment and believed she could benefit by cooperating and con‐
fessing to the murder. Owings then decided to lie to the po‐
lice, thinking that falsely admitting to the murder would
mean less jail time. She did so by concocting the story that she
and two friends were driving and using drugs when they ac‐
cidentally hit Behrman on her bicycle. Owings said that the
collision took place on Harrell Road on Bloomington’s south
side, roughly 20 miles from where Behrman’s bike was found.
To cover up the accident, Owings explained, they loaded Beh‐
rman’s body into their car, wrapped her in a plastic sheet se‐
cured with bungee cords, stabbed her, and dumped her body
in Salt Creek. Investigators were able to corroborate some of
Owings’s information: they drained the creek and found a
knife, plastic tarp, and bungee cords. Although Behrman’s
body was not recovered, the police closed the investigation
into her disappearance.
    Nearly three years after Behrman’s disappearance, in
March 2003, a father and son hunting in the woods north of
Bloomington came across a human jawbone. The woods were
about 20 miles north of where Behrman’s bike was found. The
authorities and a forensic expert surveyed the scene and col‐
lected other skeletal remains. They determined based on den‐
tal records that the remains belonged to Jill Behrman. Recog‐
nizing her story no longer added up, Owings recanted her
confession and admitted to lying about the murder in hopes
for leniency on other charges.
No. 19‐3158                                                     5

    The authorities reopened the investigation after Owings’s
recantation, but no meaningful breakthrough occurred until
2004. It was then that Detective Rick Lang turned his focus to
Myers based on unexpected information provided by Myers’s
own family. His grandmother Betty Swaffard came forward
and told the authorities that Myers had made a series of sus‐
picious and incriminating comments about Behrman’s disap‐
pearance. Others also reported incriminating statements My‐
ers made to them about the case. His former girlfriend, Carly
Goodman, likewise informed the police about a time Myers
took her to the approximate location in the woods where Beh‐
rman’s remains were later found. These developments led the
state to conclude it had enough evidence to bring charges. In
April 2006 a grand jury indicted Myers for the murder of Jill
Behrman.
   B. The Trial
       1. Opening Statements
    Trial began on October 16, 2006. In its opening statements,
the prosecution highlighted Myers’s many incriminating
statements, focusing especially on his grandmother who felt
compelled to alert the authorities despite strong feelings of
family loyalty. The state’s theory hinged on Behrman riding
her bike along a northern route on North Maple Grove near
Myers’s home, which the state said they would prove by pre‐
senting bloodhound scent evidence.
    Defense counsel opened by suggesting Myers had an alibi:
the morning that Behrman disappeared, Myers made phone
calls from the landline in his northside home at 9:15, 9:17, 9:18,
10:35, and 10:47 a.m. That timing, defense counsel suggested,
rendered Myers’s involvement impossible if Behrman rode
6                                                 No. 19‐3158

her bike not north (in the direction of Myers’s home) but in‐
stead to the south along Harrell Road. The officers involved
in the first investigation considered that route possible after
speaking to one of Behrman’s classmates and to Wendy Ow‐
ings, both of whom said they saw Behrman on that road on
the day she disappeared.
    Myers’s counsel also used his opening statement to offer
the jury two alternative suspects for the murder. The first was
Wendy Owings, the person who confessed to the murder but
later recanted her story after the police recovered Behrman’s
remains in a different place than she had identified. Defense
counsel alternatively sought to place blame on Brian Hollars,
a Bloomington resident who worked with Behrman at the Stu‐
dent Recreational Sports Center. But in contending that Hol‐
lars was responsible for Behrman’s murder, defense counsel
made certain misrepresentations. He promised the jury evi‐
dence that Hollars and Behrman were romantically involved
and were seen fighting the day before she disappeared. Coun‐
sel also represented that a bloodhound followed Behrman’s
scent in the direction of Hollars’s house but that an officer
stopped the dog before it could reach the front door. All of
those promises rang hollow, as defense counsel never pre‐
sented any such evidence.
      2. The State’s Case Against Myers
    The evidence presented during the first few days of trial
focused on how Behrman’s remains were uncovered, identi‐
fied, and analyzed. Then the state presented evidence about
her cycling habits and movements the day she disappeared.
Brian Hollars testified for the prosecution, described Beh‐
rman’s work at the recreational center, and offered an alibi by
No. 19‐3158                                                   7

informing the jury that he was at work the day of the disap‐
pearance. His testimony was not meaningfully challenged.
    As the state promised, it presented evidence supporting
its theory that Behrman rode north on North Maple Grove,
near Myers’s home. Foremost, the state presented evidence
showing the location at which Behrman’s bike was found.
Deputy Charles Douthett, who conducted a search with his
bloodhound several days after the disappearance, likewise
testified that the dog tracked Behrman’s scent along parts of
the northern route. The dog alerted to Behrman’s scent not
only in the general direction of Brian Hollars’s home but also
near the location of her bike and indeed even a touch north in
the direction of Myers’s home. The jury heard no evidence
that the bloodhound tracked Behrman to Hollars’s doorstep
as defense counsel told the jury in his opening statement.
     The state’s witnesses also included members of Myers’s
own family. His mother recounted for the jury a time in 2001
when Myers returned from fishing in the woods and reported
finding a “bone” and “panties.” Myers’s aunt Debbie Bell tes‐
tified that two months before Behrman’s disappearance, My‐
ers called asking for help watching his daughter because he
needed time alone. Bell told the jury she remembered Myers
pointing to problems with his girlfriend Carly Goodman and
saying that he “felt like he was a balloon full of hot air ready
to burst.” She also described Myers’s demeanor on the day
Behrman disappeared, recalling that he showed up at his par‐
ents’ home crying and saying he was leaving town.
   Myers’s grandmother Betty Swaffard testified despite
what she described as conflicting feelings of family loyalty.
She told the jury that early on Myers said he was a suspect in
Behrman’s disappearance and was afraid to drive past the
8                                                  No. 19‐3158

police roadblocks near his home. She further recounted My‐
ers’s statements four years later in 2004, when he called and
asked her to take care of his daughter. He explained that he
needed time to himself because he had “a lot of things” to
think about. When Swaffard asked what was wrong, Myers
said that “if the authorities knew” what he had done he
would “be in prison for the rest of [his] life.” As he dropped
off his daughter, Myers was crying and told his grandmother
that he wished he “wasn’t a bad person” and that he hadn’t
“done these bad things.” Defense counsel did not meaning‐
fully cross‐examine these family members.
   The state also presented evidence about Myers’s unusual
behavior around the time of Behrman’s disappearance. The
jury heard, for example, a neighbor explain that Myers had
covered the windows of his trailer and moved his car on the
day Behrman went missing. Myers said he parked elsewhere
so nobody could see he was home.
   Nine additional witnesses testified that Myers brought up
Behrman’s disappearance—sometimes in highly inculpatory
terms—between 2000 and 2006. One of those witnesses was
the husband of Myers’s cousin, who recalled him saying at a
family gathering in late 2001 that he bet Behrman’s body
would be found in the woods.
   Another witness, Myers’s former coworker Dean Alexan‐
der, told the jury that while out on a furniture delivery, Myers
asked him if he had heard about the Behrman case. Myers
proceeded to point out where Behrman’s bike was found and
said that he had been questioned by police a couple of times
because he lived close by. Alexander also told the jury that
Myers then went further and, while driving north, gestured
out the window and said, “if he was ever going to hide a body,
No. 19‐3158                                                   9

he would hide it up this way in a wooded area.” The state also
called Kanya Bailey, a former girlfriend of Myers, who said
that in 2000 or 2001, he pointed to the spot where Behrman’s
bike was found and told her that he was the one who found it.
      The state presented further testimony from John Roell,
who was in jail for a petty offense and shared a cell with My‐
ers for two days in May 2005. Roell recounted for the jury cer‐
tain statements Myers made about Behrman. More specifi‐
cally, Roell came forward to authorities to report that Myers
brought up the Behrman case and mentioned her bicycle three
or four times. Roell described how Myers paced nervously
about the cell, appeared to be angry, once referred to a
woman—who Roell believed was Behrman—as a “bitch,” and
said that “if she wouldn’t have said anything, this probably
. . . none of this would have happened.”
    Myers’s former girlfriend Carly Goodman also testified
for the prosecution and told the jury about the time in March
2000 when Myers drove her to a clearing in the woods north
of Bloomington. Six years later, Detective Lang drove her back
to the same general area and, without prompting, Goodman
stated that she recognized the area as the location where My‐
ers had taken her before. That area was less than one mile
from where the Morgan County hunter found Behrman’s re‐
mains. On cross‐examination Goodman acknowledged that
she had little explanation for how she recognized that clearing
compared to any other.
    Additional evidence supporting the prosecution’s theory
came from pathologist Stephen Radentz. He testified that Beh‐
rman had been killed by a shotgun wound to the back of the
head. He also opined that the physical evidence surrounding
the scene, including the failure to locate any clothing, led him
10                                                  No. 19‐3158

to conclude that Behrman was raped before she was mur‐
dered.
   A firearms expert also testified and explained that the
murder weapon, which was never recovered, likely was a 12‐
gauge shotgun. The state presented testimony from Myers’s
brother, who explained that he kept a 12‐gauge shotgun at his
parents’ house but noticed the gun was missing when he
moved home for a month in June 2000.
       3. The Defense Case
    The defense called only two witnesses. The first was Gary
Dunn, the FBI investigator who led the initial inquiry into
Behrman’s death and considered seriously the possibility that
she had biked along a southern route away from Myers’s
home. After Dunn stepped down, defense counsel admitted
being unprepared to call their next witness because they
“didn’t anticipate having to put on [their] case this early”—
the state had rested its case earlier than the trial schedule an‐
ticipated.
   Jason Fajt, an officer responsible for processing physical
evidence in the case, also testified. Fajt presented books about
pregnancy and reproductive health found in Behrman’s bed‐
room, presumably to bolster the defense’s theory that Beh‐
rman had a relationship with Brian Hollars and became preg‐
nant with his child. Fajt also showed the jury the tarp, knife,
and bungee cords found in Salt Creek that were consistent
with the confession that Wendy Owings later recanted.
       4. Closing Arguments
   The state used its summation to argue that the trial evi‐
dence exposed two of Myers’s obsessions: his ex‐girlfriend
Carly Goodman and Behrman’s bicycle. The prosecutor
No. 19‐3158                                                   11

reminded the jury of the many witnesses who described My‐
ers’s statements about Behrman’s bike. The state likewise em‐
phasized Myers’s statements to his grandmother and aunt,
urging the jury to see them as confessions to the murder.
    The state also described Myers’s activities the morning of
the murder, painting his calls to various parks and drive‐in
movies as a “last‐ditch effort to get [his girlfriend] Carly back”
and explaining that he was “trying to get control back” over
her. The state connected the two apparent obsessions by es‐
tablishing a motive: Myers wanted to control Goodman but
could not, so instead he took Behrman, who was merely in the
wrong place at the wrong time, to the same clearing in the
woods where he had driven Goodman. Based on the evi‐
dence, the state argued, Myers’s need to control women mo‐
tivated what Dr. Radentz called a “classic rape homicide.”
    On the defense side, Myers’s counsel followed up with a
watered‐down version of his original theory, since much of it
had been discredited during trial. Defense counsel reempha‐
sized Myers’s alibi and that the evidence about which way
Behrman rode was a wash. He touched on the Wendy Owings
theory and posited that the physical evidence did not rule out
a stabbing.
    When it came to Brian Hollars, counsel shied away from
his original theory. He still suggested that Behrman might
have been murdered because she was pregnant, a theory he
gleaned from books about the topic and contraception found
in her bedroom. But counsel said that the person responsible
for the murder could have been Hollars “or maybe it was an‐
other man entirely.” The defense also noted that Myers had
no clear motive for the murder and stressed the lack of phys‐
ical evidence connecting him to it.
12                                                  No. 19‐3158

    The jury deliberated for less than two hours and returned
a guilty verdict. The trial court later sentenced Myers to 65
years’ imprisonment. The Indiana Court of Appeals affirmed
Myers’s conviction and sentence on direct review. See Myers
v. State, 

(Ind. Ct. App. 2008). The Indiana Su‐
preme Court then declined review. See Myers v. State, 

(Ind. 2008) (unpublished table decision).
     C. Requests for Postconviction Relief in State Court
    Myers began his quest for postconviction relief by filing a
petition in the trial court alleging that his counsel had per‐
formed so ineffectively at trial as to violate the Sixth Amend‐
ment. To support his petition, Myers pointed to an order of
the Indiana Supreme Court finding that defense counsel, Pat‐
rick Baker, had engaged in professional misconduct during
the trial, and suspending his license to practice for six months.
See In re Baker, 

(Ind. 2011). The Indiana Su‐
preme Court found that Baker breached his ethical duties not
only by making false promises to the jury during his opening
statement, but also by improperly soliciting Myers as a client
and then falsely promising to represent him free of charge.
    While Myers alleged multiple instances of ineffective as‐
sistance in his state postconviction petition, three specific er‐
rors came to form the focus of his request for relief:
        1. False promises: Counsel’s broken promises to
           the jury in his opening statement destroyed
           his credibility and left jurors confused about
           his theory of defense.
        2. Bloodhound evidence: Counsel’s failure to ob‐
           ject to unreliable bloodhound evidence al‐
           lowed the jury to conclude that Behrman
No. 19‐3158                                                  13

          traveled near Myers’s home, providing him
          the opportunity to commit the murder.
       3. Rape testimony: Counsel failed to object to Dr.
          Radentz’s testimony that the circumstances
          around Behrman’s murder suggested she
          was raped. And that testimony allowed the
          jury to find a sexual assault motive, which
          was unfounded and resulted in severe prej‐
          udice.
    The Indiana trial court denied relief. It noted that defense
counsel did make misrepresentations during his opening
statement but said that they did not affect the trial’s outcome
because the judge instructed the jurors not to base their deci‐
sion on the statements and arguments of counsel. The trial
court also rejected all of Myers’s other contentions that his
counsel performed deficiently at trial.
   The Indiana Court of Appeals affirmed the denial of post‐
conviction relief. See Myers v. State, 

(Ind. Ct.
App. 2015). The court evaluated Myers’s three primary con‐
tentions of ineffective assistance of counsel this way:
       1. False promises: The court found deficient per‐
          formance because defense counsel knew or
          should have known that no evidence sup‐
          ported his contentions that a bloodhound
          detected Behrman’s scent at Hollars’s home.
          (The court did not address counsel’s false
          promise about evidence showing that Hol‐
          lars and Behrman were seen fighting the day
          before her disappearance.) The false promise
          about the bloodhound evidence did not
14                                               No. 19‐3158

          result in prejudice, however, as defense
          counsel was able to present some evidence
          suggesting the possibility that Hollars had a
          romantic interest in Behrman. That evi‐
          dence, the court reasoned, came from Beh‐
          rman’s cousin, who testified that some uni‐
          dentified “older man” had asked Behrman
          out on a date.
      2. Bloodhound evidence: Though defense counsel
         testified at the postconviction hearing that
         he did not remember ever researching the
         admissibility of bloodhound evidence, the
         court concluded that Myers failed to over‐
         come the presumption that counsel’s choice
         not to object to the evidence was strategic.
         Finding no deficient performance on this
         score, the court never reached the prejudice
         question.
      3. Rape testimony: The court forwent a deficient
         performance analysis and concluded that
         Baker’s failure to object to the testimony
         from Dr. Radentz did not prejudice Myers at
         trial. The court determined that counsel suc‐
         cessfully cross‐examined Dr. Radentz and
         elicited the acknowledgment that his conclu‐
         sion about rape was not based on any phys‐
         ical evidence from Behrman’s remains.
   Before denying relief, the state appellate court paused to
address Myers’s contention that counsel’s errors, when aggre‐
gated, affected the jury’s decision and thereby amounted to
ineffective assistance of counsel. The court underscored its
No. 19‐3158                                                  15

finding that Myers had failed to show even one instance of his
counsel performing deficiently in a way that resulted in prej‐
udice. Without even one error to point to, the court reasoned,
there was nothing to aggregate as part of any cumulative prej‐
udice analysis.
    The Indiana Supreme Court again declined review. See
Myers v. State, 

(Ind. 2015) (unpublished table
decision). It was then that Myers sought postconviction relief
in federal court.
   D. District Court’s Award of Federal Habeas Relief
    In a 146‐page opinion, the district court focused much of
its analysis on what it found were three serious errors com‐
mitted by Myers’s trial counsel: making false promises re‐
garding the Brian Hollars evidence during opening state‐
ments, not objecting to the bloodhound evidence, and failing
to preclude Dr. Radentz’s testimony about Behrman likely be‐
ing raped. Myers v. Superintendent, Indiana State Prison, 410 F.
Supp. 3d 958, 981, 991, 1016 (S.D. Ind. 2019).
    Turning to prejudice, the district court determined that the
Indiana Court of Appeals, as the last court to have considered
the merits of Myers’s ineffective assistance claim, considered
each allegation of ineffective assistance in isolation, rather
than focusing on their cumulative effect. See

The failure to consider the combined effect of the errors, the
district court concluded, amounted to an unreasonable appli‐
cation of the clearly established direction the Supreme Court
provided in Strickland v. Washington, 

(1984). See

district court found that “no fairminded ju‐
rist could conclude that trial counsel’s cumulative errors did
16                                                 No. 19‐3158

not meet Strickland’s prejudice standard.” 

. The court underscored that, while the
state’s evidence was sufficient to convict Myers, “it is far from
a strong case of guilt” and, as a result, “the prejudice caused
by trial counsel’s errors more likely impacted the verdict.”

Indeed, the district court found that the “cumulative
impact of trial counsel’s errors was devastating to Mr. Myers’s
defense.”

The court granted habeas relief on that
basis.
                               II
    While this appeal owes some of its complexity to the fed‐
eral habeas corpus framework, the proper starting point is fa‐
miliar. To evaluate a claim of ineffective assistance of counsel,
we apply the standard the Supreme Court announced in
Strickland v. Washington and ask whether defense counsel’s
performance was deficient and resulted in 

. The deficient performance prong requires that the de‐
fendant show that his counsel’s errors were so far below the
level of competent representation that it was as though he had
no counsel at all. See

prong, the defendant
must show “a reasonable probability that, but for counsel’s
unprofessional errors, the result of the proceeding would
have been different.”

   That hill is even steeper for issues that the state court de‐
cided on the merits. Through its enactment of the Antiterror‐
ism and Effective Death Penalty Act of 1996, Congress has al‐
lowed a federal court to award habeas relief to those like John
Myers convicted of crimes under state law only if the state
court’s ruling “was contrary to, or involved an unreasonable
application of, clearly established Federal law, as determined
by the Supreme Court,” or “was based on an unreasonable
No. 19‐3158                                                     17

determination of the facts in light of the evidence presented
in the State court proceeding.” 28 U.S.C. § 2254(d)(1), (2). This
is no easy task. See Cullen v. Pinholster, 

, 181 (2011)
(describing the federal habeas standard as “difficult to meet”
and “highly deferential”). The deferential standard reflects
Congress’s decision to require federal courts to afford sub‐
stantial respect to the interests of comity and finality embod‐
ied in state court judgments of conviction.
    By its terms, however, so‐called AEDPA deference does
not apply to federal claims that the state court did not address
on the merits. See 28 U.S.C. § 2254(d) (providing that the def‐
erential standard of review applies to “any claim that was ad‐
judicated on the merits in [s]tate court proceedings”). When a
state court reaches only one part of Strickland’s two‐pronged
analysis, we review the unaddressed prong de novo. See Wig‐
gins v. Smith, 

, 534 (2003) (reviewing de novo
whether the defendant was prejudiced for purposes of the
Strickland analysis because the state court did not reach the
issue).
   A. Deficient Performance
    We agree with the district court that defense counsel’s per‐
formance fell short of “the legal profession’s objective stand‐
ards for reasonably effective representation.” See Anderson v.
Sternes, 

, 1057 (7th Cir. 2001) (citing 

–88). We reach that conclusion whether we
evaluate counsel’s performance de novo or by affording the In‐
diana Court of Appeals’s assessment the deference prescribed
by § 2254(d)(1).
   False promises. Defense lawyers often argue for acquittals
on the basis that the authorities charged the wrong person.
18                                                   No. 19‐3158

Myers’s counsel sought to do just that but went too far. Coun‐
sel promised to present evidence that Brian Hollars killed
Behrman, even though he had to know he could not follow
through on that promise at trial. No evidence supported the
promises to prove that a bloodhound tracked Behrman’s scent
to Hollars’s home or that Hollars was seen arguing with Beh‐
rman a day or two before she disappeared. Without such evi‐
dence, counsel’s promises went unfulfilled. Making false
promises about evidence in an opening statement is a surefire
way for defense counsel to harm his credibility with the jury.
See United States ex rel. Hampton v. Leibach, 

, 257
(7th Cir. 2003). The state wisely concedes that Myers’s coun‐
sel’s false promises are indefensible—a clear instance of defi‐
cient performance.
    Bloodhound evidence. The analysis is not as straightforward
with Myers’s contention that his counsel should have objected
to the testimony that a bloodhound tracked Behrman’s scent
along a northern route and ultimately to the location of where
her bike was found. The prosecution used this evidence to put
Behrman—not just her bike—in north Bloomington, near My‐
ers’s home. Myers’s counsel may have thought that it was
more difficult to assign blame to Hollars without evidence
putting Behrman in north Bloomington the morning of her
disappearance.
    But the district court was right in its observation that coun‐
sel so passively allowing the bloodhound evidence all but
guaranteed the jury would not credit Myers’s alibi that he was
at home making telephone calls on a landline for a good part
of the morning when Behrman disappeared. And the district
court was equally correct that, at the very least, defense coun‐
sel should have investigated the admissibility of bloodhound
No. 19‐3158                                                    19

evidence and made an informed decision about whether to
seek its exclusion. We, too, are troubled by counsel’s acknowl‐
edgment at the postconviction hearing that he did not recall
doing anything to assess the admissibility of the bloodhound
evidence. Plain and simple, counsel missed the issue.
    In these circumstances, we are inclined to agree with the
district court that counsel’s failure to object amounted to de‐
ficient performance and that the Indiana Court of Appeals’s
conclusion to the contrary was unreasonable, as it assumed
without any evidentiary foundation in the record that coun‐
sel’s failure to object reflected a considered and reasonable
strategic decision. See Strick

–91 (“[S]trate‐
gic choices made after thorough investigation of law and facts
relevant to plausible options are virtually unchallengeable;
and strategic choices made after less than complete investiga‐
tion are reasonable precisely to the extent that reasonable pro‐
fessional judgments support the limitations on investiga‐
tion.”); 

(emphasizing the same point).
    Rape testimony. Perhaps the starkest example of deficient
performance came from counsel’s failure to object to testi‐
mony stating that the circumstances around the murder were
consistent with Behrman being raped before she was killed.
Dr. Radentz, a forensic pathologist, investigated Behrman’s
remains and the scene surrounding their recovery and con‐
cluded that the cause of death was a “shotgun wound to the
back of the head.” That conclusion alone presented an obsta‐
cle for Myers’s theory that Wendy Owings’s recanted confes‐
sion (that she ran over Behrman with her car, stabbed her to
death, and then dumped her body in Salt Creek) was in fact
true. But Dr. Radentz also testified that Behrman’s remains
20                                                   No. 19‐3158

being found in a remote place without any clothing suggested
that she was raped.
    In state postconviction review, the Indiana Court of Ap‐
peals did not consider whether the failure to object to Dr. Ra‐
dentz’s testimony reflected deficient performance, preferring
instead to take the permissible course of going straight to
Strickland’s prejudice prong. See 

(“[A] court need not determine whether counsel’s perfor‐
mance was deficient before examining the prejudice suffered
by the defendant as a result of the alleged deficiencies.”). That
analytical route has a consequence for our review of Myers’s
request for federal habeas relief: we lack any state court de‐
termination on the deficient performance prong to which to
review or defer under § 2254(d)(1). See Quintana v. Chandler,

, 853 (7th Cir. 2013) (“[W]hen a state court makes
the basis for its decision clear, [§] 2254(d) deference applies
only to those issues the state court explicitly addressed.”) (cit‐
ing 

). Our review therefore proceeds
de novo.
    Having taken our own fresh and thorough look at the trial
record, we conclude without hesitation that defense counsel’s
failure to object to Dr. Radentz’s testimony was “outside the
wide range of professionally competent assistance.” Strick‐

. Counsel provided no explanation for this
failure, and our role is not to search for one to excuse his de‐
ficient performance. See 

–27; Brown v.
Sternes, 

, 691 (7th Cir. 2002).
   While the Indiana Court of Appeals was right to note that
defense counsel did manage to elicit acknowledgment from
Dr. Radentz that he could not prove Behrman was raped, the
observation only goes so far. Defense counsel himself was
No. 19‐3158                                                  21

responsible for provoking the testimony most harmful to My‐
ers. On direct examination, Dr. Radentz raised only the possi‐
bility of a rape‐homicide and even then only in passing. But it
was defense counsel’s imprecise and prolonged questioning
on cross‐examination that allowed Dr. Radentz to underscore
his certainty that a rape occurred. Indeed, in response to de‐
fense counsel’s questions, Dr. Radentz testified that he consid‐
ered the case “a rape homicide and dumping until proven
otherwise.”
    Preventing the jury from hearing a word about a rape mo‐
tive should have been a priority for counsel. Everyone should
agree that the introduction of evidence of sexual violence, es‐
pecially in a case where a young college student went missing
and later turned up dead, can be prejudicial. See House v. Bell,

, 541 (2006). And in prosecuting Myers, the state
did not use Dr. Radentz’s testimony solely to explain where
and how Behrman was murdered. It instead relied on the tes‐
timony to support its theory of motive: that Myers raped Beh‐
rman before shooting her as a display of his desire to control
women. Defense counsel should have sought to prevent My‐
ers from being portrayed as a rapist.
    In the end, we agree with the district court that counsel
performed deficiently. We turn now to whether any of coun‐
sel’s errors resulted in substantial prejudice to Myers.
   B. Prejudice
    Errors are prejudicial when there is a “reasonable proba‐
bility” that the trial would have come out differently without
them. 

; see also Cook v. Foster, 

, 908 (7th Cir. 2020) (explaining and applying the
same standard). “A reasonable probability,” the Supreme
22                                                   No. 19‐3158

Court has explained, “is a probability sufficient to undermine
confidence in the outcome.” 

. While
the Supreme Court has avoided assigning a numerical proba‐
bility to the inquiry, it has explained that the likelihood of a
different result need not be “more likely than not” but none‐
theless “must be substantial.” Harrington v. Richter, 

, 111–12 (2011).
    Where, as here, the record shows more than one instance
of deficient performance, the Sixth Amendment requires that
we approach the prejudice inquiry by focusing on the cumu‐
lative effect of trial counsel’s shortcomings. This direction
comes from Strickland itself, where the Supreme Court in‐
structed courts to “consider the totality of the evidence before
the judge or 

. “Taking the unaffected
findings as a given, and taking due account of the effect of the
errors on the remaining findings, a court making the preju‐
dice inquiry must ask if the defendant has met the burden of
showing that the decision reached would reasonably likely
have been different absent the errors.”

    We have read Strickland just this way—as mandating a cu‐
mulative assessment of prejudice—on at least five prior occa‐
sions. See, e.g., Harris v. Thompson, 

, 648 (7th Cir.
2012) (“The question is whether counsel’s entire performance
at the hearing prejudiced Harris. By analyzing each deficiency
in isolation, the [state] appellate court clearly misapplied the
Strickland prejudice prong.”); Sussman v. Jenkins, 

,
360 (7th Cir. 2011) (assessing the “cumulative impact” of
counsel’s two errors); Goodman v. Bertrand, 

, 1030
(7th Cir. 2006) (reversing the district court’s denial of a § 2254
application because the state appellate court unreasonably
applied federal law by “evaluating each error in isolation”
No. 19‐3158                                                   23

and not in their totality); Washington v. Smith, 

,
634–35 (7th Cir. 2000) (explaining that the Strickland prejudice
inquiry required an assessment of “the totality of the omitted
evidence” and the other evidence presented to the jury); Alva‐
rez v. Boyd, 

, 824 (7th Cir. 2000) (explaining that
the need to analyze errors together because their “synergistic”
effects can make the “whole . . . greater than the sum of its
parts”).
    The Indiana Court of Appeals did not undertake a cumu‐
lative prejudice inquiry. It instead relied on its assessment of
each individual error in isolation and then reasoned that be‐
cause no one error met each of Strickland’s two prongs, a cu‐
mulative analysis was unnecessary. The court defended its
approach with sparse reasoning: “We have reviewed each of
Myers’ claims of error in detail and concluded that none of
them amount to ineffective assistance of counsel.” 

. It then offered the view that “trial irregulari‐
ties” cannot be combined to “gain the stature of reversible er‐
ror.”

State, 

, 1154 (Ind.
2010)). That legal observation is at odds with Strickland itself
and our prior conclusions.
    In these circumstances, where the state habeas court has
not conducted a cumulative prejudice analysis, we must un‐
dertake the inquiry on our own in the first instance. See Good‐

–31 (considering the impact of counsel’s
errors in light of the strength of the other evidence presented
to the jury de novo and therefore without deference to the state
court’s findings).
   It is here that we part ways with the district court. In eval‐
uating the state’s evidence against Myers, assessing defense
counsel’s errors, and projecting how the trial may have
24                                                   No. 19‐3158

proceeded differently absent those errors, the district court
found itself lacking confidence in the jury’s guilty verdict. The
court emphasized the absence of physical evidence linking
Myers to the murder and, even more generally, any proof of
a prior connection between him and Behrman. And, as the
district court saw it, the absence of such proof is what made
Dr. Radentz’s rape testimony so prejudicial to Myers, for it al‐
lowed the jury to assign a motive to what otherwise appeared
an implausible crime. So, too, did the district court find that
Myers suffered substantial prejudice when his counsel made
no meaningful effort to show that Behrman rode her bike
south (not north toward Myers’s home) the morning she dis‐
appeared. That evidence would have been difficult for the
prosecution to overcome, given the landline telephone rec‐
ords showing that Myers was home making calls that morn‐
ing. And that point is why, in the district court’s view, the
bloodhound scent evidence mattered: it put Behrman close to
Myers’s home and doomed his alibi.
    We see the evidence differently. Far from weak, the prose‐
cution presented substantial evidence of Myers’s guilt. The
district court failed to appreciate that, when taken together,
the evidence of Myers’s guilt overwhelmed any prejudicial ef‐
fect of defense counsel’s failings. The weight of the state’s case
against Myers prevents him from showing that he suffered
substantial prejudice from his trial counsel’s errors. See Strick‐

(“[A] verdict or conclusion only weakly
supported by the record is more likely to have been affected
by errors than one with overwhelming record support.”);

(emphasizing the same point).
   To be sure, counsel’s deficient performance undoubtedly
had some impact on the trial. The most troubling aspect of
No. 19‐3158                                                   25

counsel’s deficient performance—failing to object to Dr. Ra‐
dentz’s testimony that Behrman was raped before she was
murdered—allowed the state to supply the jury with a theory
of motive. The state made the point as plain as day in its clos‐
ing argument: “You know the motive in this crime is clear . . .
when Doctor Radentz told you that this was a classic rape
murder. Rape is a crime of control. Rape is not a sex crime. It
is pure and simple control over another human being and
dominating them.” The state used Dr. Radentz’s opinion to
underscore the narrative that Myers, while reeling from his
breakup with Carly Goodman, had a need to control people,
especially women.
    But even without the testimony about rape, the state
painted that picture about Myers through other means. The
jury heard testimony showing that Myers lost his girlfriend,
Carly Goodman, and had no luck trying to restore the rela‐
tionship, including by unexpectedly showing up at her senior
class trip and trying to join her at an amusement park in Lou‐
isville before being turned away. The jury also heard from
John Roell, who shared a cell with Myers in May 2005, that
Myers spoke about Behrman using degrading language and
saying that nothing had to happen to her if she would not
have said anything—statements evincing Myers’s attempt to
exert control over her. With all of this evidence, the state por‐
trayed a defendant who lost control of one relationship and
committed a horrific crime as part of trying to exercise control
over a young woman of a similar age.
   As for counsel’s false promises during his opening state‐
ment, we do not doubt that those damaged the theory of de‐
fense that Brian Hollars committed the murder. The jury
never heard any testimony about a bloodhound alerting to
26                                                 No. 19‐3158

Behrman’s scent near Hollars’s home or any argument be‐
tween Hollars and Behrman in the days before her disappear‐
ance.
    Counsel’s errors also weakened Myers’s alternate theory
that Wendy Owings committed the murder. Recall that when
she confessed, Owings said that she hit Behrman on Harrell
Road on the south side of Bloomington. But by not objecting
to the bloodhound evidence, counsel let the state provide sup‐
port for its theory that Behrman rode her bike in the opposite
direction the morning she went missing, making Owings’s in‐
volvement seem implausible, especially given the recanted
and admittedly false confession.
    In evaluating the whole trial picture, however, it becomes
clear that the viability of the Hollars and Owings theories was
significantly undermined for reasons other than counsel’s
mistakes. The Owings theory depended on Behrman riding
south, but apart from the bloodhound scent testimony, the
jury heard evidence that she rode in the opposite direction.
Remember that Behrman’s bike was found along North Ma‐
ple Grove, less than a mile from Myers’s home. Nothing about
the bloodhound scent testimony changed that evidentiary ob‐
stacle for Myers. And from the outset, Hollars was not a seri‐
ous suspect in the case. Indeed, the jury learned that the au‐
thorities initially considered Hollars due only to the admoni‐
tions from a psychic in Michigan.
    Much more significant, the state called both Hollars and
Owings to testify at trial. Owings explained in detail the pres‐
sure she felt from investigators and her own defense attorney
to cooperate and confess to murdering Behrman since she was
facing significant time on other charges. She also described
how she formulated the fake story, in part by relying on her
No. 19‐3158                                                27

childhood experiences swimming and fishing in Salt Creek.
For his part, Hollars denied any involvement in Behrman’s
disappearance, described the very limited interactions he had
with Behrman during her work at the Indiana University rec‐
reation center, and explained that he was at work on the cam‐
pus the morning she disappeared.
    Simply put, Myers could not compete with the testimony
the prosecution presented from Owings and Hollars—evi‐
dence presented almost certainly to prove to the jury that the
state had charged the right person with Behrman’s murder.
The testimony from Owings and Hollars diminished the
strength of Myers’s defense to a much greater extent than any
prejudice that independently followed from counsel’s failure
to object to the bloodhound scent evidence or misleading
opening statement.
   So, while we are quick to acknowledge counsel’s errors,
we are confident that the defense theories that they impacted
were sufficiently undermined, if not overwhelmed, by evi‐
dence presented at trial. Any impact from those errors on the
jury’s verdict pales in comparison to the strength of the evi‐
dence the state presented against Myers:
         Behrman’s bike was found less than a mile
          from Myers’s home;
         The very day Behrman went missing, Myers
          was seen crying and took steps to cover his
          windows with blankets and move his car to
          prevent anyone from knowing he was home;
         Myers took his girlfriend to the approximate
          location in the Morgan County woods where
28                                                   No. 19‐3158

           a hunter later came upon Behrman’s re‐
           mains;
          Myers had access to a 12‐gauge shotgun like
           the one that the experts opined was used in
           Behrman’s murder; and
          Myers made multiple self‐incriminating
           statements to many different people, with at
           least one of those statements being tanta‐
           mount to confessing to committing the mur‐
           der.
   The last point bears especially significant weight. The in‐
criminating statements Myers made to so many different peo‐
ple following Behrman’s disappearance make all the differ‐
ence in determining whether defense counsel’s errors sub‐
stantially affected the outcome of the trial. During an inter‐
view with the investigators in 2005, he insisted that had not
discussed the Behrman case with anyone except law enforce‐
ment—a position at extreme odds with much of the other tes‐
timony that the jury heard. See United States v. Rajewski, 

, 158 (7th Cir. 1975) (“It is well settled that untrue ex‐
culpatory statements may be considered as circumstantial ev‐
idence of the defendant’s consciousness of guilt.”). Myers put
himself front and center in the murder, conveying to many
people an obsession with Behrman’s disappearance and
death and thereby thwarting a meaningful chance of a suc‐
cessful defense at trial.
    The list is long, so we will recap just a few of the most re‐
vealing and inculpatory statements that Myers made. Just af‐
ter Behrman went missing, Myers called his grandmother
Betty Swaffard and asked for $200, telling her that he was a
No. 19‐3158                                                      29

suspect in the case. Around the same time, he showed up at
his parents’ home crying and said he was leaving town and
never coming back.
    Swaffard’s full testimony was devastating for Myers. She
told the jury that in 2004 her grandson called her and said that
he had “a lot of things [he] need[ed] to think about.” He then
went further and told her that if the authorities knew about
the things on his mind he would “be in prison for the rest of
[his] life.” Later that night when he dropped his daughter off
at Swaffard’s house, he cried and told her that he wished he
“hadn’t done these bad things.” Swaffard heard these state‐
ments as relating exclusively to Behrman, and, despite feel‐
ings of deep‐seated family loyalty, felt compelled to come for‐
ward and share the information with the authorities.
    Remember too that after Behrman’s disappearance, Myers
told his mother that he had been fishing in a creek in the
woods and came upon a “bone” and “panties.” He likewise
told his cousin’s husband (before the authorities recovered
Behrman’s remains) that he bet the police would find Beh‐
rman’s body in the woods.
    Aside from these statements to family members, the jury
heard from an array of friends, acquaintances, and commu‐
nity members recalling similar comments. For example, My‐
ers spoke frequently of the Behrman case and even aggran‐
dized his role in it, like falsely telling his ex‐girlfriend that he
was the one who found Behrman’s bike. Even more, he told
his former coworker Dean Alexander during a discussion
about Behrman that if he was going to hide a body, he would
hide it up north in a wooded area. Myers’s comment fore‐
shadowed what happened over a year later—Behrman’s re‐
mains were found in the woods just north of Bloomington.
30                                                   No. 19‐3158

And then there was John Roell, Myers’s former cellmate who
told the jurors that Myers brought up Behrman and her bicy‐
cle repeatedly, called her a “bitch,” and said “if she wouldn’t
have said anything, this probably . . . none of this would have
happened.”
   All of this testimony regarding the unsolicited statements
that Myers made to those around him about Behrman’s dis‐
appearance and murder went untainted by any of his trial
counsel’s errors and by any measure defeated his defense.
    Our examination of the record leaves us of the firm con‐
viction that even without counsel’s errors, the jury would
have reached the same conclusion and found John Myers
guilty of murdering Jill Behrman. Because of the strength of
the evidence presented at trial, our confidence in the jury’s
decision is not undermined. See Lee v. Avila, 

, 571
(7th Cir. 2017) (finding no prejudice despite deficient perfor‐
mance when “the state’s case was very strong” and made a
different outcome “not reasonably probable”). Myers has
fallen short of demonstrating what the Supreme Court has
told us is essential to relief rooted in a claim of ineffective as‐
sistance of counsel—that the “likelihood of a different result
must be substantial.” 

–12.
    We close by noting that the district court, while granting
Myers relief based on the three instances of ineffective assis‐
tance of counsel analyzed in this opinion, acknowledged but
did not definitively resolve other, lesser alleged instances of
ineffective assistance. Our analysis of the strength of the
state’s evidence forecloses relief based on these other allega‐
tions of ineffective assistance. But we do remand for the sole
purpose of allowing the district court to address the two
claims Myers advanced under Brady v. Maryland, 

No. 19‐3158                                                   31

(1963), in his § 2254 application. The district court reserved
judgment on these claims. Our conclusions regarding the
strength of the state’s evidence may well foreclose relief on
those claims too, but the district court should assess the ques‐
tion in the first instance as neither party briefed the claims in
this appeal.
   We REVERSE the order granting Myers’s petition for a
writ of habeas corpus and REMAND for the sole and limited
purpose of allowing the district court to consider the unre‐
solved Brady claims identified above.
32                                                          No. 19‐3158

                              Appendix




                           Map Source: Indiana State Library, Indianapolis, IN


     1. John Myers’s home
     2. Location where Jill Behrman’s bike was found
     3. Brian Hollars’s home and location where one resident thought he
        saw Behrman cycling
     4. Jill Behrman’s home
     5. Location where another resident thought she saw Behrman cycling
        and where Wendy Owings said she was driving

     Not pictured: Jill Behrman’s remains were found about 20 miles north
     of Point 2.