Rodney L. Donelson v. Troy Steele

United States Court of Appeals
For the Eighth Circuit
___________________________
No. 20-1094
___________________________
Rodney L. Donelson
lllllllllllllllllllllPetitioner - Appellant
v.
Troy Steele
lllllllllllllllllllllRespondent - Appellee
____________
Appeal from United States District Court
for the Eastern District of Missouri - St. Louis
____________
Submitted: October 22, 2021
Filed: October 26, 2021
____________
Before SMITH, Chief Judge, KELLY and ERICKSON, Circuit Judges.
____________
SMITH, Chief Judge.
A Missouri jury found Rodney Donelson guilty of two counts of first-degree
murder. After the Missouri Court of Appeals denied him postconviction relief,
Donelson filed this federal habeas action under 28 U.S.C. § 2254. Donelson claims
that he received ineffective assistance of counsel when his trial attorney withdrew a

-- 1 of 27 --

motion to sever the two murder charges. The district court1 denied his habeas petition.
We affirm.
I. Background
This case involves the murders of two women, Cassandra Scott and Barbara
Hampton, who were killed five years apart—in 2000 and 2005, respectively. In 2009,
Donelson, an acquaintance of both women, was charged with the murders. The
offenses were joined in a single indictment.
A. Pretrial Motion to Sever Offenses
Donelson’s trial counsel, Geralyn Ruess, initially filed a motion to sever the
offenses. Ruess argued that the State of Missouri had not made the requisite showing
for joinder, including any similarity in motive or character, or any common scheme
or plan. The State replied that the similarities between the two murders made joinder
proper. The state trial court held a hearing on the motion. The State called a detective
who testified about the similarities between the murders: (1) both crime scenes
involved phone cords and knives; (2) both women appeared to have had their
underwear removed; (3) both bodies were found with various liquids either on or
around them; (4) a bottle of rubbing alcohol was found at each crime scene; (5) Scott
had been strangled and Hampton was gagged and suffocated; (6) both victims were
black females; (7) both women were killed in their apartments, which were one mile
apart; (8) Donelson knew both victims; and (9) Donelson’s DNA was present at both
crime scenes.
Ruess highlighted the differences between the offenses, including: (1) the
murders occurred five years apart; (2) nothing indicated that the phone cord near
1The Honorable Audrey G. Fleissig, United States District Judge for the
Eastern District of Missouri.
-2-

-- 2 of 27 --

Scott’s body had anything to do with her murder; (3) cleaning fluid had been poured
on Hampton’s body but not Scott’s; (4) only Scott had been stabbed; and (5) only
Hampton had been sexually assaulted. Ruess also argued that prejudice from joining
the two murder charges would be “incredibly overwhelming.” Resp. to Order to Show
Cause, Ex. A, at 8, Donelson v. Steele, No. 4:16-cv-00637-AGF (E.D. Mo. 2016),
ECF No. 14-1. She explained that the DNA evidence against Donelson was weaker
in one case, but if the cases were combined, “there[] [was] no way a jury [would] be
able to give him a fair shot on both of those separate incidents.” Id.
The state trial court denied the motion, concluding that sufficient similarities
existed to justify joinder of the offenses. On the date trial began, Ruess withdrew the
motion to sever the offenses.
B. Trial
We describe the evidence from trial as summarized by the Missouri Court of
Appeals:
In July 2000, Cassandra Scott . . . was found dead in her
apartment.[2] She was lying face down on the floor in a pool of blood
and with a kitchen knife protruding from the back of her neck. A
container of antifreeze, a telephone cord, and a pair of men’s underwear
were found nearby. A purse strap was wrapped around Scott’s neck and
arm. The murderer had apparently broken a window on the front door to
gain entry into Scott’s apartment. Investigators discovered that the blood
near Scott’s body had been diluted by some other liquid and that the
liquid was on Scott’s buttocks. Investigators found an empty bottle of
2Scott was found by her boyfriend, Ronald Dickens, who testified at trial.
Dickens testified that he had been previously convicted of assault for hitting Scott
after questioning her about infidelity. He also admitted on cross-examination that he
had “used a knife to scare” Scott in the past.
-3-

-- 3 of 27 --

isopropyl alcohol in the apartment, and the knife found in Scott’s neck
matched some knives in the kitchen sink. An autopsy revealed that Scott
died from a combination of strangulation by the purse strap and five cuts
to the right side of her neck, which pierced the jugular vein. Laboratory
testing on the men’s underwear found near Scott’s body revealed that
two stains on the underwear were blood and seminal fluid. DNA tests
matched the blood to Scott and the seminal fluid to [Donelson].
[Donelson] worked at the daycare where Scott worked, and
[Donelson’s] brother lived in the apartment below Scott’s apartment.
Although police investigators questioned [Donelson] about Scott’s death
approximately two months after her body was found, [Donelson] stated
that he did not know anything about the murder. However, [Donelson]
told investigators that he had been in Scott’s apartment to repair a VCR
three days prior to her murder. [Donelson] told investigators that he
might have left a bag of clothes in Scott’s apartment, including a pair of
white boxer shorts. [Donelson] claimed that he left the clothes there
because he liked to flirt with women at the daycare center and he wanted
to look clean. [Donelson] then changed his story and said that he had
been in Scott’s apartment on the night of her murder and that they were
preparing to engage in sex when they heard a car door slam. Scott
suspected her boyfriend was there, so [Donelson] gathered his clothes,
ran down the rear stairs into his brother’s apartment, and left the
building. [Donelson’s] brother, however, denied that [Donelson] was in
his apartment on the night of the murder. When investigators confronted
[Donelson] with his brother’s denial about [Donelson’s] whereabouts,
[Donelson] subsequently changed his story again and claimed he had
been at Scott’s apartment to repair a VCR.
In September 2005, [Donelson] was living in an apartment above
the apartment of Barbara Hampton . . . . On September 14, 2005, at
approximately 10:40 p.m., Hampton was having a telephone
conversation with her daughter. Hampton interrupted the conversation
to answer a knock at the door, then told her daughter that [Donelson]
was there and wanted to use Hampton’s telephone to make a call.
Hampton ended the call with her daughter.
-4-

-- 4 of 27 --

The following day, Hampton was found dead in her apartment.
She was lying on the bedroom floor with a gag tied around her mouth.
Hampton’s dress and slip were pushed up, and her underwear had been
removed and left near her feet. Hampton had sustained an injury to her
vaginal area. Several bottles were located near Hampton’s body: dish
washing liquid, laundry detergent, and an empty bottle of isopropyl
alcohol. Near Hampton’s body, police investigators found the cap from
the bottle of isopropyl alcohol, dried liquids and powders, a kitchen
knife, and a telephone cord. An autopsy revealed that Hampton died
from suffocation caused by the gag pushing her tongue back so that it
blocked her airway. The autopsy also showed that Hampton had
sustained a fresh injury to her vaginal area that could have been caused
by a sharp object or by something stretching the tissue. Blood was found
on Hampton’s slip, on a pillowcase, and on the cap from the bottle of
isopropyl alcohol. DNA tests revealed that [Donelson] was the source
of the majority of DNA found in the blood on the bottle cap.
[Donelson’s] DNA also was found in some of the blood stains on the
pillowcase. Trace amounts of DNA consistent with [Donelson’s] DNA
was found on [the] telephone cord and on Hampton’s slip.
Police investigators questioned [Donelson] about Hampton’s
death approximately one month after her body was found. [Donelson]
told investigators that he had spoken with Hampton the night before her
body was found but that he had left with a friend named Robert Ellis . . .
and did not return home until the next morning. Ellis, however, denied
that [Donelson] had ever spent the night with him and specifically
denied that [Donelson] had spent the night with him in September
2005.[3]
Consequently, after further investigation, police questioned
[Donelson] a second time. This time, [Donelson] admitted that he had
3However, Ellis also testified that [Donelson] came over to his house “basically
every day” in September of 2005, and would stay for several hours, sometimes until
11:00 p.m.
-5-

-- 5 of 27 --

lied in his first statement because he had spent the night with Brenda
Jacobs . . . and he did not want his girlfriend, Melinda Freeman . . . , to
know he had been cheating on her. [Donelson] became angry and
agitated during the second interview with police. [Donelson] then
admitted he had been in Hampton’s apartment a few years earlier to help
her husband carry in a mattress. Hampton’s husband had died
approximately two years before Hampton’s murder. After [Donelson]
was arrested and informed of his Miranda4 rights, [Donelson] repeated
the story he had given to investigators during the second interview.
Police investigators then questioned Freeman, with whom
[Donelson] had lived in the apartment above Hampton’s apartment.
Freeman provided investigators with Jacobs’ telephone number. She
stated that [Donelson] had instructed her to tell police that he had been
cheating on her with Jacobs. Investigators subsequently interviewed
Jacobs, who stated that she and [Donelson] had not spent any night
together in September 2005.
After his arrest, [Donelson] also discussed Scott’s murder with
police investigators. [Donelson] claimed that he had gone to Scott’s
apartment the day before her murder to repair the toilet. [Donelson] said
that he must have left a bag of clothes in Scott’s apartment; however, on
the night of Scott’s murder, he had stayed with his girlfriend . . . .
State v. Donelson, 343 S.W.3d 729, 731–33 (Mo. Ct. App. 2011).
The prosecution introduced DNA evidence into the trial record. One DNA
expert, Anna Kiatowski, testified about the Scott murder. She testified that while the
DNA on the boxers matched Donelson, everything else at the crime scene matched
Ronald Dickens.
4Miranda v. Arizona, 384 U.S. 436 (1966).
-6-

-- 6 of 27 --

Two DNA experts, Samantha Webb and Malena Jimenez, testified about the
Hampton murder. Webb testified that “[e]xcluding an identical twin, Rodney
Donelson is the source of the major contributor DNA detected” on the bottle cap and
the pillow case “to a reasonable degree of scientific certainty.” Resp. to Order to
Show Cause, Ex. A, at 86. She specified that only “one in 300 quintillion” people
would have that same DNA. Id. at 87. Jimenez testified that DNA samples from
Hampton’s slip and the phone cord under Hampton’s body were “consistent” with
Donelson’s DNA. Id. at 90. Jimenez noted, however, that the specific haplotype was
more likely to be found in white individuals than black individuals: “[It] was found
0 times in 3,271 black individuals and 5 times in 3,912 Caucasian individuals . . . .
[T]he frequency of this profile [is] approximately 1 in every 1,007 total individuals,
1 in every 1,093 black individuals and 1 in every 417 Caucasian individuals.” Id.
Jimenez also testified that she had “misspelled [Donelson’s] name on [her] original
[lab] report.” Id. at 91. Her original report indicated that the DNA samples from the
Hampton crime scene matched someone named Rodney “Donaldson,” as opposed to
“Donelson.” Id. Jimenez explained that she “fix[ed] that [misspelling] and then issued
an amended report” with the correct spelling. Id.
The defense called no witnesses. During her closing argument, Ruess argued
that because “Scott is the weaker case, [the State] want[s] you to jump to Hampton
and say, ‘Oh, yeah, this is really horrible,’ because there is all this DNA and you
know that DNA never fails.” Id. at 102. She then argued that the DNA at the
Hampton crime scene did not belong to Donelson, who is black, and instead belonged
to a white man named “Donaldson.” She stated that “[n]obody compared Rodney
Donaldson’s DNA to Rodney Donelson’s DNA. D-O-N-E-L-S-O-N, not
D-O-N-A-L-D-S-O-N.” Id. She used Jimenez’s statistical analysis to suggest that the
DNA sample more likely belonged to a white person:
-7-

-- 7 of 27 --

95 percent confidence interval. One in every 1,007 total individuals, 1
in every 1,093 African-Americans, 1 in 417 Caucasians. That DNA is
more common by twice in white people than black people. Does that
mean definitively that this Rodney Donaldson is white? No. But think,
think about it.
Id. at 102–03.
The jury convicted Donelson of both counts of first-degree murder. He was
sentenced to life imprisonment without the possibility of parole for each crime.
C. Postconviction Proceedings
The Missouri Court of Appeals affirmed Donelson’s murder convictions.
Donelson moved for postconviction relief, asserting ineffective assistance of counsel
on several grounds. The only claim relevant to this appeal is that Ruess was
ineffective for withdrawing the motion to sever. At the postconviction hearing, Ruess
testified that this was part of her trial strategy because she “didn’t have very much to
work with” and her “strategy was to just point out the inconsistencies and the holes
in the state’s case, but mainly to use the different DNA results and lab results and
reports against each other to show that either they couldn’t believe one, or they
couldn’t believe both.” Resp. to Order to Show Cause, Ex. K, at 5–6, Donelson v.
Steele, No. 4:16-cv-00637-AGF (E.D. Mo. 2016), ECF No. 14-11. She said the
inconsistencies were that
[o]ne was spelled Donaldson with an A-L-D-S-O-N, and I argued that
was not the right person. And the other one, the DNA that they found,
the frequency in which it would be found in a population was higher in
a white population than an African-American population. So I argued
that that actually would be the wrong person, because obviously it
would occur more often in a white population.
-8-

-- 8 of 27 --

Id. at 10–11. Ruess explained that after the motion to sever was denied, she later
withdrew it because she “needed both sets of DNA to order and kind of contrast
against each other” and she “thought that the judge was going to grant it, and by that
point, [she] wouldn’t have had any defense at all if he had done that.” Id. at 15.
The postconviction court denied Donelson’s petition, holding that Ruess’s
performance was not constitutionally deficient under Strickland5 because she gave a
“plausible strategic reason” for her actions:
[S]he withdrew the motion because she was concerned that the Court
would reconsider and the motion would be granted. She said she wanted
to compare and contrast the DNA evidence from the two murders. The
name on the lab report for one case did not have [Donelson’s] name
spelled properly and she wanted to be able to argue it was not the correct
person. The DNA from the other case, from the pillow case, was more
likely to be a white person than a black person. She said that because
[Donelson’s] alibi did not work out, her best defense was to attack the
DNA evidence and she believed this strategy would be more effective
if the cases were tried together.
Donelson v. Missouri, No. 0922-CR03577, 2013 WL 9768487, at *1 (Mo. Cir. Ct.
Sept. 17, 2013) (footnote omitted).
The Missouri Court of Appeals affirmed. It concluded that Ruess had a
“defensible trial strategy” of comparing and contrasting the DNA because “[t]he name
on the lab report for one case had been spelled incorrectly and the sample from the
other case was determined to be more likely to have come from a Caucasian, which
Appellant was not.” Resp. to Order to Show Cause, Ex. J, at 8, Donelson v. Steele,
No. 4:16-cv-00637-AGF (E.D. Mo. 2016), ECF No. 14-10. “Because severing the
5Strickland v. Washington, 466 U.S. 668 (1984).
-9-

-- 9 of 27 --

offenses would prohibit [her] from employing her reasonable defense strategy,” it
concluded, she was not ineffective for withdrawing her motion to sever. Id.
D. Habeas Corpus Petition
Donelson filed a federal habeas petition under 28 U.S.C. § 2254, alleging
ineffective assistance of counsel. The district court denied relief, but it granted a
certificate of appealability on the severance issue. It called it a “close question”
whether Ruess was ineffective for withdrawing the motion. Donelson v. Steele, No.
4:16-cv-00637-AGF, 2019 WL 4750280, at *9 (E.D. Mo. Sept. 30, 2019). The court
found counsel’s explanation “troubling” for several reasons: (1) Donelson “admitted
to being in both apartments and provided at least some explanation of the presence
of his DNA there”; (2) even if Ruess believed the best strategy was to argue that the
DNA did not belong to Donelson, “it [was] not clear to the [c]ourt why she needed
to join the murder charges to do so”; (3) and “trial counsel never actually ‘compar[ed]
or contrast[ed]’ the DNA evidence from the two crime scenes at trial.” Id. (third and
fourth alterations in original). The district court noted that “the state court’s
suggestion that name-error was in one case, and the race issue (Caucasian versus
African-American) was in the other appear[ed] to be a mischaracterization of the
evidence of trial.” Id. at *9 n.11 (citation omitted).
Despite “its significant concerns,” the district court gave the state appellate
court the benefit of the doubt pursuant to “the ‘doubly deferential’ standard of
review” for § 2254 claims. Id. at *9 (citing Fenstermaker v. Halvorson, 920 F.3d 536,
540 (8th Cir. 2019)). Emphasizing Ruess’s “difficult task” and “resist[ing] the
temptation to second-guess [counsel’s] trial strategy,” the district court concluded that
Donelson had not shown that Ruess’s choice to try the charges together “so clearly
outweighed her alternative options that it was objectively unreasonable for the state
court to find that [she] acted within the ‘wide range’ of reasonable professional
-10-

-- 10 of 27 --

assistance under Strickland.” Id. at *10 (second alteration in original) (quoting
Blackmon v. White, 825 F.2d 1263, 1265 (8th Cir. 1987)). Donelson timely appealed.
II. Discussion
Donelson argues that the district court erred in concluding that he was not
entitled to habeas relief on the grounds of ineffective assistance of counsel. “On
appeal from a district court’s ruling on a habeas petition, ‘we review the district
court’s findings of fact for clear error, and its conclusions of law de novo.’” Ali v.
Roy, 950 F.3d 572, 574 (8th Cir. 2020) (quoting Escobedo v. Lund, 760 F.3d 863, 868
(8th Cir. 2014)).
The Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) provides
that when, as here, a claim has been adjudicated by a state court, habeas relief is only
permissible if the state court’s determination either (1) “resulted in a decision that was
contrary to, or involved an unreasonable application of, clearly established Federal
law”; or (2) “was based on an unreasonable determination of the facts in light of the
evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d). AEDPA
raises a high bar to relief: “a state prisoner must show that the state court’s ruling on
the claim being presented in federal court was so lacking in justification that there
was an error well understood and comprehended in existing law beyond any
possibility for fairminded disagreement.” Harrington v. Richter, 562 U.S. 86, 103
(2011). This standard, which “stops short of imposing a complete bar on federal-court
relitigation of claims already rejected in state proceedings,” is “difficult to meet . . .
because it was meant to be.” Id. at 102. “We will not lightly conclude that a State’s
criminal justice system has experienced the ‘extreme malfunctio[n]’ for which federal
habeas relief is the remedy.” Burt v. Titlow, 571 U.S. 12, 19–20 (2013) (alteration in
original) (quoting Richter, 562 U.S. at 102).
-11-

-- 11 of 27 --

We review claims of ineffective assistance of counsel under Strickland’s
two-part test, “which requires [a petitioner] to show that counsel’s performance fell
below professional norms and that, as a result of that deficient performance, he was
prejudiced.” Gabaree v. Steele, 792 F.3d 991, 996 (8th Cir. 2015). Thus, “[t]o grant
relief under § 2254, . . . we must conclude that the state court unreasonably applied
the Strickland test or that, in reaching its conclusion regarding the performance of [a
petitioner’s] attorney, it made unreasonable factual conclusions.” Id. If a petitioner
overcomes § 2254’s high bar, then we are not limited by AEDPA deference and
review the issue de novo. Id. at 999.
Donelson contends that the state appellate court’s decision was both
(1) contrary to, or involved an unreasonable application of Strickland, and (2) based
on an unreasonable determination of the facts. Because we agree that the state court’s
decision was based on an unreasonable determination of the facts, we address only
the second prong.
A. Unreasonable Determination of the Facts
“The question under AEDPA is not whether a federal court believes the state
court’s determination was incorrect but whether that determination was
unreasonable—a substantially higher threshold.” Schriro v. Landrigan, 550 U.S. 465,
473 (2007). “We may not characterize . . . state-court factual determinations as
unreasonable ‘merely because [we] would have reached a different conclusion in the
first instance.’” Brumfield v. Cain, 576 U.S. 305, 313–14 (2015) (second alteration
in original) (quoting Wood v. Allen, 558 U.S. 290, 301 (2010)). Rather, we may
“supersede the trial court’s determination” only “[i]f reasonable minds reviewing the
record [would not] disagree about the finding in question.” Id. at 314 (cleaned up);
see also Bahtuoh v. Smith, 855 F.3d 868, 873 (8th Cir. 2017) (“A state court decision
is based on an unreasonable determination of the facts only if the court’s
presumptively correct factual findings do not enjoy support in the record.” (quotation
-12-

-- 12 of 27 --

omitted)). This is a high bar, but “‘[e]ven in the context of federal habeas, deference
does not imply abandonment or abdication of judicial review,’ and ‘does not by
definition preclude relief.’” Brumfield, 576 U.S. at 314 (alteration in original)
(quoting Miller-El v. Cockrell, 537 U.S. 322, 340 (2003)).
Donelson argues that the state court based its decision on two unreasonable
determinations of fact: (1) it erroneously found that the two inconsistencies identified
by Ruess—the name-error issue and the race issue—occurred in different cases, and
(2) it erroneously attributed to Ruess a strategy the record shows she did not follow.
We agree.
First, it is undisputed that the Missouri Court of Appeals made an incorrect
factual statement when assessing Ruess’s performance. The court wrote:
Counsel had decided her defense strategy at trial would be to compare
and contrast the DNA samples taken from each of the two crime scenes.
The name on the lab report for one case had been spelled incorrectly
and the sample from the other case was determined to be more likely to
have come from a Caucasian, which Appellant was not. Accordingly,
counsel’s defense strategy was to argue the DNA from the murder
scenes did not belong to Appellant.
Because severing the offenses would prohibit counsel from
employing her reasonable defense strategy, withdrawing the motion to
sever was not ineffective.
Resp. to Order to Show Cause, Ex. J, at 8 (emphasis added). The second, highlighted
statement is wrong; both of those issues existed in the Hampton case, and neither
existed in the Scott case. One expert witness—Jimenez—discussed both of those
issues, and she only testified about the Hampton case. In closing, Ruess highlighted
-13-

-- 13 of 27 --

the name error and race issue to argue that the DNA in Hampton’s case—not
Scott’s—more likely belonged to a white person named Donaldson.
“Because [the government] concedes that . . . th[is] factual statement[] [was]
erroneous, we will assume that [Donelson] has overcome the presumption of [its]
correctness by clear and convincing evidence.” Collier v. Norris, 485 F.3d 415, 423
(8th Cir. 2007) (citing 28 U.S.C. § 2254(e)(1)). Still, “it does not necessarily follow
that the state court adjudication was based on an unreasonable determination of facts
because subsection (d)(2) instructs federal courts to evaluate the reasonableness of
the state court decision ‘in light of the evidence presented in the State court
proceeding.’” Id. (quoting 28 U.S.C. § 2254(d)(2)). Here, however, we conclude that
the state court’s erroneous finding was unreasonable because the trial transcript
makes the error abundantly clear. “[R]easonable minds reviewing the record [would
not] disagree.” Brumfield, 576 U.S. at 314 (quoting Wood, 558 U.S. at 301).
Our inquiry does not end there, however. “[I]t is not sufficient to show the state
court’s decision merely included an unreasonable factual determination. Instead, by
its terms § 2254(d)(2) only empowers federal courts to grant relief if the state court’s
decision was based on an unreasonable determination of the facts.” Smith v. Aldridge,
904 F.3d 874, 880 (10th Cir. 2018) (quotation omitted). The government urges that
the state court did not base its conclusion on this erroneous fact because it relied on
“three facts,” only one of which was wrong. Appellee’s Br. at 38. “[T]he two other
facts,” it argues, “enjoy support in the record.” Id. at 39.
We are not persuaded that the presence of two technically accurate statements
effectively cancels out a third, inaccurate one. Rather, we agree with Donelson that
“the ‘three factual bases’ for the [state] court’s decision were not separate and
independent from one another.” Appellant’s Reply Br. at 12. This is indicated by the
state court’s use of the word “[a]ccordingly” to transition from the second sentence
-14-

-- 14 of 27 --

to the third. See, e.g., The Redbook: A Manual on Legal Style, R. 10.40, at 207 (3d.
ed. 2013) (“A conjunctive adverb joins two clauses and indicates the relationship
between them . . . .”). True, Ruess said that her strategy was to compare and contrast
the DNA evidence to argue that the DNA was not Donelson’s, but the state court’s
understanding of how or why Ruess would have pursued that strategy appears to rest
on the second sentence—i.e., the incorrect one. As Donelson puts it, “[t]he Missouri
court’s finding that ‘trial counsel’s strategy was to compare and contrast the lab
reports’ was completely dependent on finding the second fact that the court
found—that there were problems with the DNA in both cases.” Appellant’s Reply Br.
at 12–13. Although we cannot say precisely how much it relied on the erroneous fact,
it is hard to imagine that the state court’s reasoning and conclusion would have been
the same had it understood that in fact, those two issues occurred in only one of the
cases.
Second, we agree with Donelson that the state court made an unreasonable
determination of fact when it explained that Ruess’s strategy was to “compare and
contrast the DNA samples taken from each of the two crime scenes.” Resp. to Order
to Show Cause, Ex. J, at 8. “We cannot impute to counsel a trial strategy that the
record reveals she did not follow.” Gabaree, 792 F.3d at 999. We also cannot accept
a “state court’s determination of counsel’s strategic decisions [that] ‘resembles more
a post hoc rationalization of counsel’s conduct than an accurate description of [her]
deliberations.’” Id. (second alteration in original) (quoting Wiggins v. Smith, 539 U.S.
510, 526–27 (2003)). If counsel’s purported strategy has no support in the record and
the court’s reliance on it amounts to a “post hoc rationalization,” this results in an
unreasonable determination of the facts. Id.
Here, the state court’s explanation of Ruess’s strategy amounted to a post-hoc
rationalization. We agree with the district court’s assessment that
-15-

-- 15 of 27 --

trial counsel never actually “compar[ed] or contrast[ed]” the DNA
evidence from the two crime scenes at trial. Rather, her discussion of the
DNA evidence at trial consisted of arguing that some of the DNA
evidence in Hampton’s case (but not in Scott’s case) may have belonged
to an unknown white man named Rodney Donaldson, rather than
[Donelson], and stating in closing argument that “nobody” had
compared the DNA from the two cases.
Donelson, 2019 WL 4750280, at *9 (first and second alterations in original). Indeed,
the only link she drew between the DNA in each case was to repeatedly comment that
nobody had bothered to check whether the DNA sample labeled “Donaldson” was in
fact Donelson’s. Moreover, we have no trouble finding that the state court based its
decision on this post-hoc rationalization because it relied on this purported strategy
to conclude that Ruess was not ineffective.
B. Ineffective Assistance of Counsel Claim
Because the state court’s conclusion that Ruess performed effectively relied on
unreasonable determinations of fact, Ҥ 2254(d) does not require that we adhere to
that court’s decision on this issue, and our review is de novo.” Gabaree, 792 F.3d at
999. Our review of Strickland’s prejudice prong is also de novo because the state
court did not assess prejudice, and we are thus unconstrained by § 2254. Id. We move
straight to the question of prejudice because “[f]ailure to establish either Strickland
prong is fatal to an ineffective-assistance claim,” and “we determine that no prejudice
resulted from counsel’s alleged deficiencies.” Ramirez v. United States, 751 F.3d 604,
607 (8th Cir. 2014) (first quoting Worthington v. Roper, 631 F.3d 487, 498 (8th Cir.
2011); then quoting Gianakos v. United States, 560 F.3d 817, 821 (8th Cir. 2009));
see also Strickland, 466 U.S. at 697 (“If it is easier to dispose of an ineffectiveness
claim on the ground of lack of sufficient prejudice, . . . that course should be
followed.”).
-16-

-- 16 of 27 --

To show prejudice, “[t]he defendant must show that there is a reasonable
probability that, but for counsel’s unprofessional errors, the result of the proceeding
would have been different. A reasonable probability is a probability sufficient to
undermine confidence in the outcome.” Strickland, 466 U.S. at 694. This standard is
less onerous than the preponderance-of-the-evidence standard. See id. The likelihood
of a different outcome, however, “must be substantial, not just conceivable.” Richter,
562 U.S. at 112. Donelson argues that had Ruess not withdrawn the motion to sever,
there is a reasonable probability that (1) the cases would have been severed, either on
reconsideration by the trial court or on appeal; and (2) once severed, he would have
received a different verdict in one or both cases. Both these contentions must be true
for him to show a reasonable probability of a different outcome. We assess each
contention in turn.
1. Likelihood of Severance on Appeal
To assess the likelihood of severance, our inquiry focuses on Donelson’s claim
that the charges would have been severed on appeal.6 In Missouri,
Appellate review of a claim for failure to sever charges involves
a two-step analysis. First is a determination of whether the initial joinder
of the offenses was proper; and, second, upon a determination that the
joinder was proper, this court must then determine whether the trial
court abused its discretion in refusing to sever the offenses. Where
offenses are improperly joined, prejudice is presumed from a joint trial
and the trial court erred in overruling the motion for severance. . . .
6Donelson has not shown any probability that the trial court would have
reconsidered and granted the severance motion on its own. The court had already
denied the motion and, aside from Ruess’s testimony that she thought the judge might
grant it, there is no evidence suggesting that the trial court was going to change its
mind.
-17-

-- 17 of 27 --

Whether joinder is proper or improper is a question of law, while
severance is within the trial court’s discretion. Joinder addresses the
more basic question of what crimes can be charged in a single
proceeding, while severance presupposes proper joinder and leaves to
the trial court’s discretion the determination of whether prejudice may
or would result if charges properly joined were tried together.
State v. Saucy, 164 S.W.3d 523, 528 (Mo. Ct. App. 2005) (quotations and citations
omitted). Donelson thus has the burden of convincing us that there exists a reasonable
probability that the Missouri Court of Appeals would have reversed on either the
joinder or severance issue. This means he must show that joinder was improper as a
matter of law or that the trial court abused its discretion in denying severance.
Missouri favors “[l]iberal joinder of criminal offenses.” State v. McKinney, 314
S.W.3d 339, 341 (Mo. 2010) (en banc). Joinder is proper “if the offenses charged . . .
are of the same or similar character.” Id. (alteration in original) (quoting Mo. Rev.
Stat. § 545.140(2)); see also Mo. Sup. Ct. R. 23.05. “The tactics used in separate
offenses need not be identical to constitute acts of the same or similar character.”
State v. McMillon, 436 S.W.3d 663, 670 (Mo. Ct. App. 2014). However, “the manner
in which the crimes were committed [must] be sufficiently similar that it is likely that
the same person committed all the charged offenses.” State v. Hughes, 787 S.W.2d
802, 804 (Mo. Ct. App. 1990). Missouri courts “may consider several non-exclusive
factors, including the commission of the same type of offenses, victims of the same
sex and age group, offenses occurring at the same location, or offenses closely related
chronologically.” McMillon, 436 S.W.3d at 670.
Here, several similarities supported joinder: (1) the charges were both for first-
degree murder; (2) the victims were both black females; (3) not only did both females
know Donelson, but his brother was Scott’s neighbor and Donelson was Hampton’s
neighbor; (4) the crime scenes were one mile apart; (5) both crime scenes involved
-18-

-- 18 of 27 --

a phone cord, knife, and an empty bottle of rubbing alcohol; (6) a bottle of antifreeze
was near Scott’s body, and bottles of cleaning liquids and powders were near
Hampton’s body; (7) Scott’s blood had been diluted with liquid, and Hampton’s body
was covered in cleaning liquids; (8) both women were found without underwear; and
(9) both murders involved strangulation or suffocation.
We acknowledge that there were also important differences. For example,
(1) there was no evidence that the phone cord near Scott’s body had anything to do
with her murder; (2) Scott was stabbed in the neck, whereas the knife in Hampton’s
case was found a few feet from her body; (3) Hampton experienced injuries to the
vaginal area, indicating sexual assault, whereas there was no evidence that Scott was
sexually assaulted; and (4) significant time—five years—elapsed between the crimes.
However, “tactics need only resemble or correspond with one another” for joinder to
be proper, State v. Holliday, 231 S.W.3d 287, 293 (Mo. Ct. App. 2007), and we find
that Missouri’s non-exclusive factors favor joinder. We need not determine with
absolute certainty that joinder was proper. To meet his burden, Donelson must show
more than that there is a chance that the Missouri Court of Appeals would have found
joinder improper on appeal. He has not.
Donelson faces an even higher hurdle regarding severance because he must
show that the trial court abused its discretion. The denial of a motion to sever “will
be reversed if the trial court abused its discretion in overruling the motion and if there
was a clear showing of prejudice.” McKinney, 314 S.W.3d at 342. Abuse of discretion
occurs when the trial court’s “ruling is clearly against the logic of the circumstances
then before the court and is so arbitrary and unreasonable as to shock the sense of
justice and indicate a lack of careful consideration.” Id.
Severance is proper only after the defendant “‘makes a particularized showing
of substantial prejudice if the offense is not tried separately’ and . . . the ‘court finds
-19-

-- 19 of 27 --

the existence of a bias or discrimination against the party that requires a separate trial
of the offense.’” Id. (quoting Mo. Sup. Ct. R. 24.07)). Courts consider factors such
as “the number of offenses charged, the complexity of the evidence to be offered and
whether the trier of fact can realistically distinguish the evidence and apply the law
intelligently to each offense.” State v. Sims, 764 S.W.2d 692, 697 (Mo. Ct. App.
1988). “The general allegation that the jury would likely consider evidence of guilt
on one charge as evidence of guilt on another charge does not meet the requirement
of a particularized showing of substantial prejudice.” State v. Simmons, 158 S.W.3d
901, 909 (Mo. Ct. App. 2005) (quotation omitted). “If evidence relating to each
offense is distinct and uncomplicated and the jury is properly instructed to return
separate verdicts for each offense charged, the trial court does not abuse its discretion
in denying a motion to sever.” State v. Tolen, 304 S.W.3d 229, 236 (Mo. Ct. App.
2009).
Here, the trial court’s ruling to deny severance was reasonable and did not
amount to an abuse of discretion. There were two murders, each with distinct
evidence. While there was some scientific discussion of DNA evidence, the evidence
presented was not particularly complex.7 The jury could have intelligently
distinguished the evidence in each case and applied the law to each offense.
Moreover, the jury was properly instructed that “[e]ach count must be considered
separately” and that it “should return a separate verdict for each count.” Resp. to
Order to Show Cause, Ex. B, at 36, Donelson v. Steele, No. 4:16-cv-00637-AGF
(E.D. Mo. 2016), ECF No. 14-2. Because he did not show particularized prejudice,
we find it unlikely that the Missouri Court of Appeals would have reversed the trial
court’s denial of severance.
7Donelson points to the state court’s erroneous statement of fact to support his
argument that the evidence was confusing to the jury. However, we find that the state
court’s error most likely stemmed from Ruess’s own testimony at the postconviction
hearing. As discussed, the trial transcript makes the error obvious.
-20-

-- 20 of 27 --

2. Likelihood of a Different Verdict in Either Case
Even if Donelson could show a substantial probability of severance on appeal,
he must still show an overall reasonable probability of a different outcome in the
challenged proceeding. See Strickland, 466 U.S. at 694. “When a defendant
challenges a conviction, the question is whether there is a reasonable probability that,
absent the errors, the factfinder would have had a reasonable doubt respecting guilt.”
Hinton v. Alabama, 571 U.S. 263, 275 (2014) (quoting Strickland, 466 U.S. at 695).
Because there were two guilty verdicts, we assess whether there is a substantial
likelihood of a different outcome in either case.
The Scott case was the weaker of the two, but not so weak that it would have
resulted in a different outcome if tried alone. Scott had a tumultuous and violent
relationship with her boyfriend. Dickens, the boyfriend, had abused her in the past.
Donelson’s DNA was found only on the boxer shorts. But those boxer shorts, with
his DNA and Scott’s blood on them, were found close to Scott’s body. Donelson
knew Scott, had access to her building, and admitted to being in her apartment around
the time of the murder. He changed his story multiple times about why and when he
was there—first stating that he was there several days earlier to fix her VCR; then
explaining that he was there the night of the murder to have sex with Scott but left
through his brother’s apartment; and then reverting back to the story about the VCR.
Donelson’s credibility was further undermined by his brother’s testimony that
Donelson never came through his apartment that night. While it is certainly possible
that the Scott case would have seemed weaker to a jury without the Hampton case
attached to it, we see no reasonable probability of a different verdict.
There is even less of a probability of a different outcome in the Hampton case,
where Donelson’s DNA was in several significant places. One DNA expert testified
that “[e]xcluding an identical twin, Rodney Donelson is the source of the major
contributor DNA detected” on the bottle cap and pillow case near Hampton’s body,
and that only “[o]ne [person] in 300 quintillion” would share the same DNA. Resp.
-21-

-- 21 of 27 --

to Order to Show Cause, Ex. A, at 86, 87. The other DNA expert testified that the
DNA on the phone cord and Hampton’s underwear were consistent with Donelson’s.
Donelson lived upstairs from Hampton, and Hampton’s daughter testified that
Hampton said Donelson was at her door shortly before the murder. During the
investigation, Donelson changed his story multiple times in this case as well. Further,
the two people with whom he claimed to have spent the evening of Hampton’s
murder—Ellis, then Jacobs—denied being with him.
In sum, the evidence against Donelson was convincingly incriminating in both
cases. His DNA was found at both crime scenes, there was evidence that he was in
both apartments on the nights of the murders, and he lied to police and changed his
story multiple times in both cases. Donelson has not met his burden of showing a
reasonable probability of a different outcome in either case, even if they had been
severed.
III. Conclusion
We conclude that although Donelson has demonstrated that the state court’s
decision was based on an unreasonable determination of the facts, his ineffective
assistance of counsel claim fails because he cannot show prejudice. While it may be
“conceivable” that the state appellate court would have granted severance on appeal
and that one juror would have had a reasonable doubt regarding the Scott case, the
probability is not “substantial.” Richter, 562 U.S. at 112.
For the foregoing reasons, we affirm the district court’s decision.8
KELLY, Circuit Judge, dissenting.
8While this appeal was pending, Donelson filed a pro se motion to present new
evidence. Donelson’s motion is denied.
-22-

-- 22 of 27 --

I agree that the state court’s decision was based on an unreasonable
determination of the facts. However, I dissent from the court’s determination that
Donelson has failed to demonstrate he was prejudiced by his trial counsel’s
withdrawal of the motion to sever the two murder charges. See Strickland v.
Washington, 466 U.S. 668, 687 (1984) (“First, the defendant must show that
counsel’s performance was deficient. . . . Second, the defendant must show that the
deficient performance prejudiced the defense.”).
I think there is a reasonable probability9 that, had defense counsel not
withdrawn the motion to sever before trial, the Missouri Court of Appeals would have
found joinder improper and severed the charges on appeal. In concluding otherwise,
the court today overstates the similarities between the murders. Cf. State v. Saucy,
164 S.W.3d 523, 528-29 (Mo. Ct. App. 2005) (explaining that “[t]he use of similar
tactics in the commission of multiple crimes is sufficient to show that the offenses are
of the same or similar character” only when “[t]he manner in which the crimes were
committed [was] so similar that it is likely that the same person committed all the
charged offenses”). First, Dr. Michael Graham, St. Louis’s chief medical examiner,
testified that strangulation and suffocation are “quite different” methods. The
physical action used to suffocate someone, blocking the flow of air to their lungs, is
necessarily different than the one used for strangulation, which compresses the veins
in a person’s neck. Second, the presence of common household items at both crime
scenes is not a meaningful similarity, at least not in the role they played in the
murders or their coverups. The liquid around Scott’s body was water, and though
there was a bottle of antifreeze nearby and an empty bottle of rubbing alcohol in the
9It is important to note that the standard is only “reasonable probability.”
Strickland, 466 U.S. at 694. Though “[t]he likelihood of a different result must be
substantial,” Harrington v. Richter, 562 U.S. 86, 112 (2011), the Supreme Court has
indicated that this can be something less than 50%. See id. at 111-12 (explaining that
the standard “does not require a showing that counsel’s actions more likely than not
altered the outcome” (cleaned up)).
-23-

-- 23 of 27 --

apartment, the prosecution’s expert testified that Scott’s blood did not test positive
for any other substance. And there was nothing to suggest that Scott’s killer tried to
use any of these liquids to destroy evidence. Hampton’s body, on the other hand, was
covered in multiple chemicals and detergents that were intended, according to the
prosecution, to “dilute any evidence that was there.” Third, it is unclear how the fact
that both women were found without underwear in their own home is indicative of
a similar method of committing murder. Scott did not have underwear on but none
was found near her body and there was no evidence that she had been sexually
assaulted. Hampton’s underwear, by contrast, was at her feet and she had been
sexually assaulted before her death. On this record, the “similarities” the court has
listed as favoring joinder are largely superficial.
Focusing on the “manner in which the crimes were committed,” State v.
Hughes, 787 S.W.2d 802, 804 (Mo. Ct. App. 1990), it also becomes apparent that the
differences between the two murders are significant. Scott was beaten and stabbed
multiple times in the neck, and her killer also strangled her with a purse strap.
Hampton, on the other hand, was sexually assaulted and died from a gag in her mouth
after her tongue blocked her airway. The crime scenes further demonstrate the stark
differences between the murders. A window was broken in Scott’s apartment,
whereas Hampton’s showed no signs of forced entry. Scott’s body was surrounded
by a huge pool of blood, whereas Hampton’s murder, while obviously violent, left
only a few blood stains behind—likely from the sexual assault. Scott was also found
lying face-down, whereas Hampton was on her back with her knees up. And
significantly, five years separated the two murders. Compare State v. Scott, 548
S.W.3d 351, 361 (Mo. Ct. App. 2018) (finding joinder was proper where “all three
robberies were closely related in time, occurring over a three week period”), with
State v. Bird, 1 S.W.3d 62, 63, 67 (Mo. Ct. App. 1999) (finding joinder improper
where at least seven months separated a burglary from the receipt of stolen property).
This gap in time coupled with the evidence presented at trial shows that the methods
and means used to commit the crimes do not “resemble or correspond with one
-24-

-- 24 of 27 --

another.” State v. Holliday, 231 S.W.3d 287, 293 (Mo. Ct. App. 2007); cf. Saucy,
164 S.W.3d at 529 (finding that offenses were not of the same or similar character
where “[t]he evidence did not reflect that the same weapon was necessarily used, nor
was there anything distinctive about the manner of committing the crimes”); State v.
Kelly, 956 S.W.2d 922, 925-26 (Mo. Ct. App. 1997) (concluding that offenses were
improperly joined where there was “nothing distinctive about the tactics used which
made it likely that same person was involved in each robbery”). Donelson has shown
a reasonable probability that the Missouri Court of Appeals would have found joinder
improper as a matter of law.
Donelson has also demonstrated a reasonable probability that the Missouri
Court of Appeals would have found the trial court’s failure to sever the cases an
abuse of discretion. “Where offenses are improperly joined, ‘prejudice is presumed
from a joint trial . . . .’” Saucy, 164 S.W.3d at 528 (quoting State v. Bechhold, 65
S.W.3d 591, 594 (Mo. Ct. App. 2002)). But “[e]ven where joinder is proper, . . .
severance may be necessary to prevent substantial prejudice to the defendant that
could result if the charges are not tried separately.” State v. McKinney, 314 S.W.3d
339, 342 (Mo. banc 2010).
Here, though the jury was properly instructed to separately consider the counts,
“the evidence relating to each offense” was not “distinct and uncomplicated.” State
v. Tolen, 304 S.W.3d 229, 236 (Mo. Ct. App. 2009). For one, the evidence included
highly technical genetic and statistical analyses involving two different types of DNA
testing, PCR and Y-chromosome, which in turn produce different types of results.
Compare Trial Tr. at 355:2-4 (DNA analyst testifying that Y-chromosome testing of
phone cord resulted in DNA profile consistent with both Reginald Donelson and
Rodney Donelson), with id. at 236:14-16 (DNA analyst who used PCR testing
testifying that DNA in seminal fluid “was the DNA of Rodney Donelson”). The jury
heard testimony that some DNA conclusively belonged to Donelson—such as the
seminal fluid from the Scott crime scene—but at other points heard about DNA that
-25-

-- 25 of 27 --

was merely consistent with his profile (and also, in some instances, consistent with
that of other individuals), creating a significant risk of confusion as to the proper
interpretation of the various DNA evidence. Further complicating matters was the
presence of co-mingled DNA at both crime scenes, as well as trace amounts of DNA
that could not be identified at all. In short, the DNA evidence required particularized
and nuanced interpretation, but it involved multiple, technical variables that were
difficult to keep track of. Moreover, the testimony on the two murders was not
entirely compartmentalized. Two of the state’s witnesses—Dr. Graham and Detective
Heather Sabin—testified about both of the murders, making it that much harder to
distinguish the evidence. The prosecution itself had a hard time differentiating
between the murders, describing in the opening statement to the jury “the slip that
Cassandra was wearing,” id. at 156:11-12, even though the evidence shows it was
Hampton, not Scott, who was wearing a slip. The complexity and muddled nature of
the evidence made the possibility of juror confusion—and in turn, the risk of
prejudice—substantial.10 On this record, I believe there is a reasonable probability
that the Missouri Court of Appeals would have severed the counts on appeal.
Finally, I believe Donelson has also shown a reasonable probability that, had
the counts been severed, the jury would have had reasonable doubt regarding his
guilt—at least with respect to the Scott murder. See Strickland, 466 U.S. at 695
(“When a defendant challenges a conviction, the question is whether there is a
reasonable probability that, absent the errors, the factfinder would have had a
reasonable doubt respecting guilt.”). The circumstantial evidence implicating Ronald
Dickens, Scott’s boyfriend, cannot be understated. There is testimony that (1)
Dickens had physically abused Scott in the past and threatened her with a knife; (2)
Dickens lied to police by telling them Scott had died by suicide; (3) the building
10The state appeals court itself failed to “distinguish the evidence and apply the
law intelligently to each offense.” State v. Sims, 764 S.W.2d 692, 697 (Mo. Ct. App.
1988). It is reasonable to expect the jury to have done the same.
-26-

-- 26 of 27 --

manager may have heard Scott and Dickens fighting a few hours before police
arrived; (4) Dickens’s car was seen in the parking lot outside Scott’s apartment the
night of the murder; (5) the building manager, who also worked with Scott at a
daycare center, testified that Scott never had company over other than Dickens; (6)
Dickens’s DNA was found at Scott’s apartment; and (7) Scott’s blood was found on
the living room light switch and Dickens told police that he turned the living room
light on when he first went into the apartment and found Scott’s body. Without the
prejudicial inferences against Donelson that a double murder trial would inevitably
create, there is a substantial likelihood that the evidence implicating Dickens would
have provided reasonable doubt in the jury’s mind as to Donelson’s guilt in the Scott
murder.
For these reasons, I respectfully dissent.
______________________________
-27-

-- 27 of 27 --