NOT RECOMMENDED FOR PUBLICATION
File Name: 17a0433n.06
No. 16-2336
UNITED STATES COURT OF APPEALS
FOR THE SIXTH CIRCUIT FILED
Jul 24, 2017
ROBERT ANTHONY BOJAJ, ) DEBORAH S. HUNT, Clerk
)
Petitioner-Appellant, )
)
v. ) ON APPEAL FROM THE
) UNITED STATES DISTRICT
MARY BERGHUIS, Warden, ) COURT FOR THE EASTERN
) DISTRICT OF MICHIGAN
Respondent-Appellee. )
)
)
BEFORE: GIBBONS, ROGERS, and DONALD, Circuit Judges.
ROGERS, Circuit Judge. Robert Bojaj was convicted of second-degree murder in
Michigan after he rear-ended another car while intoxicated, killing the driver of the rear-ended
car. The state trial court allowed the prosecution to introduce data from an event data recorder in
Bojaj’s car at trial without first examining the reliability of the evidence, despite Bojaj’s request
for a Daubert hearing on the issue. Bojaj now seeks a writ of habeas corpus, arguing that the
state trial court’s error denied him due process by rendering his trial fundamentally unfair.
Because the Michigan Court of Appeals’ decision denying Bojaj’s constitutional claim on the
merits was neither contrary to nor an unreasonable application of clearly established Supreme
Court precedent, Bojaj is not entitled to relief.
While intoxicated, Bojaj rear-ended another car with his Lexus sedan on a divided
highway at 1:30 a.m. on August 15, 2010, killing the woman who was driving the rear-ended car.
No. 16-2336, Bojaj v. Berghuis
People v. Bojaj, No. 303884, 2012 WL 12941199, at *2 (Mich. Ct. App. Nov. 27, 2012). A jury
convicted Bojaj of second-degree murder for the incident in a Michigan trial court. Id. at *1. At
trial, Bojaj did not dispute that he was driving while intoxicated when he caused the wreck. Id.
His blood alcohol measured .26, three times over the legal limit of .08, two hours after the crash.
The Government’s expert testified that this amounted to at least ten drinks for a man of Bojaj’s
size. However, Bojaj did dispute that he had the requisite mens rea for second-degree murder,
id., which in Michigan requires the defendant to have acted with at least “a high probability that
[his actions would] result in death and . . . with a base antisocial motive and with wanton
disregard for human life,” People v. Goecke, 579 N.W.2d 868, 880 (Mich. 1998).
The prosecution sought to prove the requisite intent through lay and expert testimony
about Bojaj’s erratic driving and speed immediately before the wreck. Four eyewitnesses
testified about Bojaj’s driving. Two, a married couple who were riding down the highway
together when Bojaj passed them, estimated that Bojaj was traveling 100 mph. The couple also
observed Bojaj hit the rumble strips on the highway’s left shoulder for three to five seconds
before quickly swerving across three lanes of traffic to the rumble strips on the right side of the
highway. Bojaj, 2012 WL 12941199, at *2. Another witness “described the Lexus as traveling
‘at a very high rate of speed’ and ‘swerving.’” Id. The fourth eyewitness did not estimate
Bojaj’s speed but saw Bojaj’s car swerving from the left to the right side of the highway and
hitting the rumble strips. Id.
The prosecution’s remaining evidence involved data from an event data record (EDR) in
Bojaj’s Lexus and an expert crash re-constructionist’s evaluation using that data. The Michigan
Court of Appeals described this evidence in detail:
After establishing these facts, the prosecutor presented the testimony of Mark
Jakstis, a Design and Technical Analysis Manager at Toyota Motor Sales, U.S.A.
-2-
No. 16-2336, Bojaj v. Berghuis
Jakstis described the Lexus’s EDR as “part of the airbag system,” with the
capability of recording certain information during a “trigger event” such as a
crash. The EDR stores information acquired milliseconds before an accident,
including the vehicle’s speed, the engine speed, the accelerator pedal position, and
the driver’s brake use. The last vehicle speed recorded by the EDR in defendant’s
Lexus was 78.3 mph, which Jakstis described as “the maximum recordable speed
for this vehicle.” The EDR reported that the accelerator was “in the mid
position,” which meant that the driver was “pushing down on the accelerator a
bit” and “still accelerating” at the time the EDR stopped recording data.
According to Jakstis, the EDR reported that the Lexus’s brakes had not been
applied.
In addition to this information, Jakstis claimed that the EDR contained the
Lexus’s “Delta–V,” which he defined as “the change in velocity of the vehicle
during the crash.” Jakstis interpreted the delta-V data as consistent with a
“34 mile an hour change in speed during this crash,” adding “and it’s still going
up.” Under cross-examination Jakstis admitted that Toyota considered the EDR
data system in defendant’s Lexus a “prototype tool,” and that the specific device
was used only in Toyota vehicles. He conceded that Toyota designed the EDR
“to look at the performance of an airbag system in a crash” rather than to aid law
enforcement after an accident.
Michigan State Police Sergeant Kevin Lucidi, a “traffic crash reconstructionist,”
utilized the EDR data to formulate one of the two speed determinations he
provided to the jury. Lucidi first calculated defendant’s speed at the time of the
impact without resorting to the EDR data, instead basing his calculations on the
“conservation of linear momentum formula.” This approach relies on “basic
physics princip[le]s” predicated on physically-measurable components including
the coefficient of friction, vehicle weights, the distance the vehicles traveled post-
impact, and the speed of the victim’s vehicle. Because the speed of [the victim’s]
vehicle at the time of impact was unknown, Lucidi assigned it three different
speeds: 50, 60 and 70 mph. After plugging these numbers into an equation,
Lucidi calculated speed ranges for the Lexus. He opined that based on this
mathematical formula, the Lexus’ minimum speed at impact ranged from 82 to 98
mph and that its maximum speed fell between 109 and 126 mph.
Lucidi then described a second, independent methodology for calculating
defendant’s speed which incorporated the delta-V supplied by the EDR.
Applying the delta-V number testified to by Jakstis, Lucidi opined that if the
[victim] was traveling between 50 and 60 mph at the time of impact, the Lexus’s
maximum speed ranged from 117 to 126 mph. Thus, Lucidi concluded that the
top number in the speed range—126 mph—was the same regardless of the
formula he employed.
The defense countered Lucidi’s opinions with the testimony of accident
reconstructionist Larry Petersen. Petersen criticized several of the assumptions
-3-
No. 16-2336, Bojaj v. Berghuis
built into Lucidi’s calculations. For example, Petersen estimated that the front
end of the Lexus contacted 50% of the rear of the [victim’s car], while Lucidi
believed the overlap percentage to have been 80%. Petersen believed that the
correct coefficient of friction differed from the figure Lucidi relied on, because
the pavement likely was somewhat damp. Petersen described the manner in
which these variables changed the ultimate speed calculations. In Petersen’s
view, defendant’s vehicle was moving at a speed between 85 and 97 mph when
the crash occurred.
Id. at *2–*3.
Before trial, Bojaj had requested a Daubert hearing to challenge the foundation of
Lucidi’s accident-reconstruction testimony, which the trial court had denied.1 After the jury
returned a guilty verdict on the second-degree murder charge, Bojaj moved for a new trial,
arguing that the trial court’s admitting the EDR data without holding a Daubert hearing to rule
on the evidence’s reliability violated Bojaj’s due process rights by denying him a fair trial. Id. at
*3–*4. The trial court agreed and granted Bojaj’s request for a new trial. Id. In so ruling, the
court explained that before trial it did not understand the nature of Bojaj’s Daubert request but
that the EDR’s reliability was seriously questioned at trial—the read-out tool used to extract
information from the EDR is a prototype, Toyota warns that EDR data is not reliable unless
independently corroborated, and the EDR was developed to provide airbag statistics, not for
accident reconstruction. The court also noted: “This certainly isn’t some harmless error, as
well. . . . It’s not curable.”
However, the Michigan Court of Appeals reversed the trial court’s decision. See Bojaj,
2012 WL 12941199, at *5. The state appellate court agreed that, “before allowing Lucidi to use
the data as a foundation for his testimony, the trial court was required to hold a Daubert
hearing.” Id. Although the court recognized that Bojaj had alleged a due process violation, it
1
Michigan Rule of Evidence 702 incorporates the reliability standards announced by the Supreme Court in Daubert
v. Merrell Dow Pharm., Inc., 509 U.S. 579 (1993). Gilbert v. DaimlerChrysler Corp., 685 N.W.2d 391, 408 (Mich.
2004).
-4-
No. 16-2336, Bojaj v. Berghuis
labelled the trial court’s error “preserved nonconstitutional error.” Id. at *4, *5. The court of
appeals then explained that the error did not warrant a new trial because it was harmless:
Even assuming that the EDR data was erroneously admitted, the exceedingly high
speed at which defendant was driving immediately before the accident was
independently corroborated by eyewitnesses and Lucidi’s conventional speed
calculation. Admittedly, the delta-V and other evidence purportedly derived from
the EDR carried the weight of advanced technology. But other powerful evidence
demonstrated that defendant elected to drive after consuming an enormous
amount of alcohol, operated his vehicle at a speed approximating 100 mph,
repeatedly crossed multiple traffic lanes, and ignored the rumble strip’s audible
warning that he was out of control. Given this evidence, any impropriety in
admitting the EDR data did not undermine the reliability of the verdict.
Id. at *5. The Michigan Supreme Court denied Bojaj’s application for leave to appeal this
decision. See People v. Bojaj, No. 146805, 834 N.W.2d 502 (Mich. 2013).
Bojaj then petitioned the district court for a writ of habeas corpus, arguing that the trial
court’s error violated the Fourteenth Amendment’s Due Process Clause by depriving him of a
fundamentally fair trial. The district court denied the petition. See Bojaj v. Berghuis, No. 14-
CV-12193, 2016 WL 4502438, at *9 (E.D. Mich. Aug. 29, 2016). The district court began its
analysis by examining whether the Michigan Court of Appeals adjudicated Bojaj’s due process
claim on the merits, which would subject Bojaj’s petition to AEDPA’s deferential standard of
review. See id. at *4. Although the district court left the question undecided, it indicated that de
novo review was likely appropriate:
[I]t appears that the Michigan Court of Appeals simply cast the failure to hold a
Daubert hearing as “nonconstitutional” error without ever saying why there was
no due-process violation . . . . Perhaps an argument could be made that the
Michigan Court of Appeals had—silently—concluded that the absence of a
Daubert hearing “qualifie[d]” as nonconstitutional error because that error was
not severe enough to rise to a constitutional due-process violation. Although the
Court thinks this is a somewhat strained reading of the appellate court’s opinion,
it is not an implausible one.
Id. (second alteration in original).
-5-
No. 16-2336, Bojaj v. Berghuis
The district court then held, applying de novo review, that Bojaj’s due process argument
fails. The court cited Ege v. Yukins, 485 F.3d 364 (6th Cir. 2007), for the relevant constitutional
standard: “Whether the admission of prejudicial evidence constitutes a denial of fundamental
fairness turns upon whether the evidence is material in the sense of a crucial, critical highly
significant factor.” Bojaj, 2016 WL 4502438, at *4 (quoting Ege, 485 F.3d at 375). The trial
court’s admitting the EDR data without verifying its reliability did not create a constitutional
violation under this standard, the court reasoned, because even if the data were inadmissible, “a
number of factors . . . substantially reduced the prejudice of its admission.” Id. at *5. “First, the
jury heard how the delta-v might be unreliable.” Id. “Second, Bojaj was able to counter Lucidi
with his own expert.” Id. “Third, expert testimony was not the only evidence the jury heard
about Bojaj’s excessive speed. At least three eyewitnesses thought that Bojaj was driving in the
range of 100 mph.” Id. The district court also noted that excessive speed was not the only factor
present to prove the intent required for second-degree murder; the prosecutor argued to the jury
that Bojaj’s extreme level of intoxication and erratic driving, in addition to his speed, supported
the requisite mens rea. Id. Therefore, the district court held, because the EDR data was not so
prejudicial as to deprive Bojaj of a fundamentally fair trial, no constitutional violation occurred,
and there is no grounds for habeas relief. Bojaj now appeals.
We review Bojaj’s due process claim under AEDPA deference, because the Michigan
Court of Appeals adjudicated the claim on the merits. Bojaj undoubtedly presented a due
process argument to the Michigan Court of Appeals, which recognized the constitutional
argument in its opinion then expressly treated the issue as “preserved nonconstitutional error,”
Bojaj, 2012 WL 12941199, at *4, *5. In such a situation, when a state court was faced with a
habeas petitioner’s constitutional argument but silently rejected it, “a federal habeas court must
-6-
No. 16-2336, Bojaj v. Berghuis
presume” that the state court adjudicated the federal claim on the merits. Johnson v. Williams,
133 S. Ct. 1088, 1096 (2013); see Harrington v. Richter, 562 U.S. 86, 99–100 (2011). Bojaj
makes no argument to rebut the strong Harrington–Johnson presumption. Thus, the Michigan
Court of Appeals’ decision is subject to AEDPA deference under 28 U.S.C. § 2254(d).
Applying AEDPA deference, the Michigan Court of Appeals did not err in its merits
adjudication that the trial court’s admitting the EDR data without holding a Daubert hearing did
not implicate the Due Process Clause. As Judge Boggs explained in a portion of his dissent in
Ege, “we have observed on a number of occasions that when a habeas claim is predicated on
evidentiary issues, relief depends on the existence of [Supreme Court] precedent establishing the
particular type of evidence at issue as violating the defendant’s due process rights.” Ege, 485
F.3d at 383 (Boggs, J., dissenting). For example, in Bugh v. Mitchell, 329 F.3d 496 (6th Cir.
2003), a petitioner argued that the state trial court had deprived him of a fundamentally fair trial
by admitting propensity evidence against him in the form of prior bad acts. We denied the
petitioner’s claim under AEDPA deference, reasoning that “no clearly established Supreme
Court precedent . . . holds that a state violates due process by permitting propensity evidence in
the form of other bad acts evidence.” Id. at 512; see also Maldonado v. Wilson, 416 F.3d 470
(6th Cir. 2005) (admission of polygraph results). In Bugh, how the propensity evidence affected
the petitioner’s trial did not matter to our AEDPA analysis. Here, we similarly focus AEDPA
deference on whether the trial court’s particular state-law evidentiary error implicates the
Constitution, not on how the error affected Bojaj’s trial.
Because the Michigan Court of Appeals’ decision that the trial court’s error did not
render Bojaj’s trial fundamentally unfair is neither contrary to nor involves an unreasonable
application of clearly established Supreme Court precedent, Bojaj is not entitled to habeas relief.
-7-
No. 16-2336, Bojaj v. Berghuis
The Supreme Court has never held that admitting evidence of questionable reliability against a
criminal defendant without holding a Daubert hearing or similar procedure violates due process.
Furthermore, it was not objectively unreasonable for the Michigan Court of Appeals effectively
to decide that this type of error does not implicate due process. The Supreme Court did not set a
constitutional floor for the reliability of scientific evidence in Daubert, see Norris v. Schotten,
146 F.3d 314, 335 (6th Cir. 1998), and screening evidence through Daubert’s standards is not
constitutionally required. Rather, as the district court explained, “state rules of evidence are the
primary safeguard against unreliable expert testimony being put before the jury[,] and the Due
Process Clause serves as a distant backstop to these rules.” Bojaj, 2016 WL 4502438, at *6; see
Dowling v. United States, 493 U.S. 342, 352 (1990). Thus, the Michigan Court of Appeals was
not unreasonable to treat the trial court’s error in admitting the EDR data as solely a non-
constitutional issue.
Even applying the case-specific approach from Ege that the district court employed,
Bojaj is not entitled to habeas relief. In Ege, we evaluated whether a criminal defendant was
denied due process by a state court’s erroneous admission of prejudicial evidence in a nuanced
way, by analyzing the extent to which the challenged evidence prejudiced the defendant’s
criminal trial. 485 F.3d 364; see Brown v. O’Dea, 227 F.3d 642 (6th Cir. 2000) (taking the same
approach). Ege explained that, if the challenged evidence, contrasted against the prosecution’s
properly admitted evidence and the defendant’s rebuttal evidence, was “a crucial, critical highly
significant factor” in bringing about a guilty verdict, the improperly admitted evidence creates a
due process violation. See Ege, 485 F.3d at 375. The dissenting opinion in Ege described that
the majority’s analysis was “at odds” with how this court had previously approached similar due
process challenges in Bugh and Maldonado. See id. at 383 (Boggs, J., dissenting). We need not
-8-
No. 16-2336, Bojaj v. Berghuis
decide whether Ege marks a change in this court’s approach to analyzing when a state-law
evidentiary error creates a due process violation, however, because even following Ege, the trial
court’s error did not render Bojaj’s trial fundamentally unfair. Even assuming the EDR data and
Lucidi’s related testimony would have been excluded under Daubert, the evidence’s prejudice
was tempered for all the reasons listed by the district court: the jury heard about the data’s
unreliability, Bojaj rebutted Lucidi with his own accident reconstructionist, lay-witness
testimony confirmed that Bojaj was driving recklessly and speeding near 100 mph, and factors
other than Bojaj’s speed support the jury’s finding of the requisite mens rea for second-degree
murder. Thus, under either a case-specific or categorical analysis, the Michigan Court of
Appeals did not err by denying Bojaj’s due process claim.
Finally, Bojaj makes a strange argument that the doctrine of invited error precludes the
warden from arguing against his habeas petition. The argument clearly fails. Bojaj says that,
because the government had once argued that a Daubert hearing was not required, the
government could not later defend the position that, if a Daubert hearing was required, denial of
the hearing was harmless error. The invited-error doctrine prohibits a party from requesting a
particular position, and then later claiming that the same position is error. See Harvis v.
Roadway Express, Inc., 923 F.2d 59, 61 (6th Cir. 1991). That did not happen here. Because the
government’s stance in this case has consistently been that no Daubert hearing was required and
that the EDR data was admissible, the doctrine does not apply. Nothing prevented the warden
from arguing that, even if the district court erred, the error either did not amount to a
constitutional violation or was harmless.
The judgment of the district court is affirmed.
-9-