State v. Juan J. Castillo

COURT OF APPEALS
DECISION NOTICE
DATED AND FILED This opinion is subject to further editing. If
published, the official version will appear in
the bound volume of the Official Reports.
June 29, 2021
A party may file with the Supreme Court a
Sheila T. Reiff petition to review an adverse decision by the
Clerk of Court of Appeals Court of Appeals. See WIS. STAT. § 808.10
and RULE 809.62.

Appeal No. 2020AP983-CR Cir. Ct. No. 2017CF832

STATE OF WISCONSIN IN COURT OF APPEALS
DISTRICT III

STATE OF WISCONSIN,

PLAINTIFF-RESPONDENT,

V.

JUAN J. CASTILLO,

DEFENDANT-APPELLANT.

APPEAL from a judgment of the circuit court for Outagamie
County: JOHN A. DES JARDINS, Judge. Reversed and cause remanded for
further proceedings.

Before Stark, P.J., Hruz and Seidl, JJ.

¶1 STARK, P.J. Juan Castillo appeals a judgment convicting him of
one count of first-degree sexual assault of a child (sexual contact with a person
No. 2020AP983-CR

under the age of thirteen). See WIS. STAT. § 948.02(1)(e) (2019-20).1 Castillo
argues the circuit court erred by excluding his expert witness’s testimony
regarding factors that can affect the reliability of a child’s statements. Castillo
also argues the court erred by denying his motions for a mistrial after two
witnesses made statements during their testimony that violated the court’s pretrial
evidentiary rulings.

¶2 We reject Castillo’s argument that the circuit court erroneously
exercised its discretion by excluding his expert witness’s testimony. The court
reasonably concluded that the proffered testimony was inadmissible because it was
not sufficiently tied to the facts of the case and was likely to confuse the jury.

¶3 We agree with Castillo, however, that the circuit court erroneously
exercised its discretion by denying his motions for a mistrial. In doing so, we
acknowledge that when a defendant’s request for a mistrial is not based on any
laxness or overreaching by the prosecution, we must give great deference to the
circuit court’s ruling. See State v. Bunch, 191 Wis. 2d 501, 507, 529 N.W.2d 923
(Ct. App. 1995). We also acknowledge that the court instructed the jury to
disregard the statements in question, and that we generally presume jurors follow
the court’s instructions. See State v. Truax, 151 Wis. 2d 354, 362, 444 N.W.2d
432 (Ct. App. 1989). Nevertheless, we conclude that in this case, the combined
prejudicial effect of the relevant statements was so great that the court’s cautionary
instructions were insufficient to remedy the error. As Castillo aptly states, under
the circumstances present here, the “evidentiary bell” simply “could not be

1
All references to the Wisconsin Statutes are to the 2019-20 version unless otherwise
noted.

2
No. 2020AP983-CR

unrung” after the jury heard the statements in question. The court therefore erred
by refusing to grant Castillo a mistrial. Accordingly, we reverse Castillo’s
judgment of conviction and remand for a new trial.2

BACKGROUND

¶4 On December 2, 2017, the State filed a criminal complaint charging
Castillo with one count of first-degree sexual assault a child (sexual intercourse
with a person under age twelve), contrary to WIS. STAT. § 948.02(1)(b). The State
later amended the charge to first-degree sexual assault of a child (sexual contact
with a person under age thirteen), contrary to § 948.02(1)(e).

¶5 The complaint alleged that Castillo had sexually assaulted his cousin
Gail3 on or about June 1, 2016. Gail was five years old at the time of the alleged
assault, and Castillo was sixteen. According to the complaint, Gail’s mother,
Anne, reported the assault to police in August 2017. Gail was examined at a
hospital, and approximately one month later she participated in a recorded forensic
interview.4

2
Because we reverse on the grounds that the circuit court erred by refusing to grant
Castillo a mistrial, it is not strictly necessary for us to address Castillo’s argument that the court
erred by excluding his expert witness’s proffered testimony. See Turner v. Taylor, 2003 WI App
256, ¶1 n.1, 268 Wis. 2d 628, 673 N.W.2d 716 (court of appeals need not address all issues raised
by the parties if one is dispositive of the appeal). We choose to do so in the interest of judicial
economy, however, because the admissibility of the expert’s testimony is likely to arise again on
remand. See State v. Temby, 108 Wis. 2d 521, 527, 322 N.W.2d 522 (Ct. App. 1982).
3
Pursuant to the policy underlying WIS. STAT. RULE 809.86, we refer to the victim, her
mother, and her sister using pseudonyms.
4
Anne also told police that Castillo had sexually assaulted her other daughter, Dana.
Dana participated in a forensic interview, but she did not disclose any sexual assault during that
interview.

3
No. 2020AP983-CR

¶6 The complaint alleged that during the summer of 2016, Gail, her
sister Dana, and Castillo were all staying at Castillo’s mother’s home. During her
forensic interview, Gail stated that one night while she was sleeping, Castillo
picked her up from the bed and carried her into his room. He then told her to
“suck it” and clarified that he wanted her to “suck his private part.” Gail “closed
her eyes and did what he told her to do.” Gail also reported that Castillo put his
“private part” in her “butt” while she was lying down on her side with her pants
off.

¶7 Before trial, Castillo filed several motions in limine. As relevant
here, the circuit court granted Castillo’s motions to exclude evidence
that: (1) Castillo had a history of being arrested, being the subject of warrants,
being on probation or parole, being on electronic monitoring, being incarcerated,
or being convicted of specific crimes; (2) Castillo was currently serving or had
previously served a sentence as a result of a criminal conviction; and (3) Castillo
had engaged in any illegal activity not alleged in the criminal complaint, including
any claim that he had sexually assaulted Dana. Citing the rape shield statute, WIS.
STAT. § 972.11(2), the court denied Castillo’s motion to introduce evidence that
Gail had allegedly engaged in “sexual play activity” with a male child before
accusing Castillo of sexual assault.

¶8 Castillo also sought permission before trial to introduce the expert
testimony of Dr. David Thompson, a clinical and forensic psychologist.
Thompson had reviewed the recording and transcript of Gail’s forensic interview
and authored a written report containing his findings. In the report, Thompson
identified six factors that “extensive research has shown affect[] the reliability of a
child’s statements”: repeated interviewing; external influences; inappropriate
interviewing techniques; interviewer bias; therapy effects; and source

4
No. 2020AP983-CR

misattribution errors. Thompson’s report then discussed how those factors may
have affected the reliability of Gail’s statements regarding the alleged sexual
assault by Castillo.

¶9 The State objected to Thompson’s proposed testimony, arguing it
was inadmissible under WIS. STAT. § 907.02, which was amended in 2011 to adopt
the federal standard for the admissibility of expert testimony set forth in Daubert
v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993). See Seifert v.
Balink, 2017 WI 2, ¶6, 372 Wis. 2d 525, 888 N.W.2d 816. The State stipulated
that Thompson was qualified as an expert witness. The State argued, however,
that Thompson’s proposed testimony was inadmissible because his report
contained only “generalized statements that are unrelated to the facts of the case
and assumption[s] not borne out by the evidence.” The State also asserted that
Thompson’s testimony would usurp the role of the jury by improperly
commenting on the believability of a witness, in violation of State v. Haseltine,
120 Wis. 2d 92, 352 N.W.2d 673 (Ct. App. 1984).

¶10 The circuit court held a Daubert hearing, during which Thompson
testified consistent with the contents of his report.5 Following the hearing, the
court denied Castillo’s request to admit Thompson’s testimony at trial for three
reasons. First, the court concluded Thompson’s proposed testimony was not
sufficiently reliable under WIS. STAT. § 907.02 and Daubert because Thompson
was merely extrapolating from general principles without any specific evidence
that the factors he discussed applied to Gail. Second, the court concluded that

5
We discuss the substance of Thompson’s report and testimony in greater detail below.
See infra, ¶¶28-37.

5
No. 2020AP983-CR

without any evidence tying Thompson’s testimony to the specific facts of this
case, his testimony was speculative and “a lot of his analysis would be confusing
to the jury.” Third, the court stated Thompson’s testimony “would have a
tendency to … interfere with the jury being the primary determiner of credibility
of the witnesses,” in violation of Haseltine.

¶11 The case then proceeded to a jury trial, at which both sides presented
various witnesses. Ultimately, however, there was no physical evidence of the
alleged assault, nor was there any witness who directly corroborated either Gail’s
or Castillo’s version of events. As such, the crucial issue at trial was whether the
jurors believed Gail or Castillo to be more credible.

¶12 Gail—who was then eight years old—testified on the first day of
trial. Castillo asserts (and the State does not dispute) that her testimony regarding
the alleged assault was generally consistent with her forensic interview.6 At the
end of Gail’s direct examination, the prosecutor asked her, “Is there anything else
that you remember about what happened that you think we should know?” Gail
responded, “He did it to three other little girls.” The prosecutor immediately
asked the circuit court to strike Gail’s response. The court granted that request,
stating, “That will be struck and the jury is ordered to disregard that comment.”

¶13 Following Gail’s testimony, Castillo moved for a mistrial, arguing
her statement that Castillo “did it to three other little girls” was so prejudicial that

6
Gail’s forensic interview was not played for the jury. Instead, a police officer testified
about what Gail said during that interview. Castillo’s attorney also read portions of the interview
transcript into the record during his cross-examination of Gail. On appeal, Castillo notes that
during the forensic interview, Gail stated she was asleep before the assault, and Castillo picked
her up from a bed and carried her into his room. In contrast, Gail testified at trial that Castillo
told her to come into his room before the assault.

6
No. 2020AP983-CR

the circuit court could not “simply unring the bell and have the jury not remember
that or instruct them not to consider that.” The court denied Castillo’s motion,
reasoning that the prosecutor’s question did not elicit Gail’s response, that the jury
would “understand[] that a child might blurt something out,” and that the jurors
were capable of following the court’s instruction to disregard Gail’s comment.

¶14 Castillo testified in his own defense. He denied ever having sexual
contact with Gail, taking off his clothes in her presence, or taking off her clothes.
He also denied ever sleeping in the same room as Gail or being alone with her.
Castillo admitted that he had fifteen prior juvenile delinquency adjudications.

¶15 Castillo’s sister, Lorena Castillo, also testified for the defense. On
cross-examination, Lorena admitted that she spoke to Castillo often, including on
the night before trial. She also admitted that she was aware her phone calls with
Castillo were recorded. When asked whether she talked to Castillo about
testifying at his trial, she replied, “Not really. We—when I—when it’s our visits,
we talk about what we want out of this.” When the prosecutor subsequently asked
what she and Castillo talked about, Lorena responded: “We talked about him
getting a job if he gets out.” Castillo’s counsel then objected, and the court stated,
“[T]he jury should disregard the part about getting out and I’ll strike that from the
record.”

¶16 Castillo again moved for a mistrial, asserting it was “pretty explicit
that [Lorena] was talking to [Castillo] when [Castillo] was in custody.” Castillo’s
attorney subsequently conceded, however, that Lorena’s comment about Castillo
“get[ting] out” “could have been” interpreted in a number of different ways. The
circuit court denied Castillo’s second mistrial motion, noting that Lorena’s
comment was not specifically elicited by the prosecutor’s question, that Lorena

7
No. 2020AP983-CR

was a hostile witness toward the prosecution, and that the court had immediately
instructed the jury to disregard her remark.

¶17 After the close of evidence, the circuit court instructed the jury,
among other things, to disregard all stricken testimony. The court also instructed
the jury that evidence of witnesses’ prior crimes could be used only to evaluate
their truthfulness. In addition, the court specifically instructed the jury that
evidence regarding Castillo’s prior juvenile delinquency adjudications could be
used only to evaluate his credibility, could not be used for any other purpose, and
was not “proof of guilt of the offense now charged.”

¶18 The jury returned a guilty verdict after deliberating for less than one
hour. The circuit court later sentenced Castillo to five years’ initial confinement,
followed by five years’ extended supervision. Castillo now appeals, arguing that
the court erred by excluding his proffered expert testimony and denying his
requests for a mistrial.

DISCUSSION

I. Expert testimony

¶19 We review a circuit court’s decision to admit or exclude expert
testimony under the erroneous exercise of discretion standard of review. State v.
Chitwood, 2016 WI App 36, ¶30, 369 Wis. 2d 132, 879 N.W.2d 786. We will
therefore uphold the court’s decision as long as it examined the relevant facts,
applied a proper legal standard, and used a demonstrated rational process to reach
a reasonable conclusion. Id. This standard is “highly deferential” to the circuit
court. Id. (citation omitted). “The question is not whether we would have

8
No. 2020AP983-CR

permitted the evidence to come in or whether we agree with the circuit court’s
ruling, but whether, in fact, appropriate discretion was exercised.” Id.

¶20 The admission of expert testimony is governed by WIS. STAT.
§ 907.02. Prior to 2011, § 907.02 provided that expert testimony was admissible
as long as the expert witness was qualified, the evidence would assist the trier of
fact, and the evidence was relevant. State v. Dobbs, 2020 WI 64, ¶33, 392 Wis. 2d
505, 945 N.W.2d 609. Our supreme court interpreted the pre-2011 version of
§ 907.02 as allowing the admission of expert exposition testimony—i.e., an expert
“testifying in the form of an educational lecture on general principles,” without
specifically applying those principles to the facts of the case at bar or stating any
opinion regarding that case. Dobbs, 392 Wis. 2d 505, ¶¶31, 33.

¶21 In 2011, the legislature amended WIS. STAT. § 907.02 to adopt the
federal reliability standard articulated in Daubert. Dobbs, 392 Wis. 2d 505, ¶34.
The statute, as amended, provides:

If scientific, technical, or other specialized knowledge will
assist the trier of fact to understand the evidence or to
determine a fact in issue, a witness qualified as an expert by
knowledge, skill, experience, training, or education, may
testify thereto in the form of an opinion or otherwise, if the
testimony is based upon sufficient facts or data, the
testimony is the product of reliable principles and methods,
and the witness has applied the principles and methods
reliably to the facts of the case.

Sec. 907.02(1). Under this standard, the circuit court “stands as a gatekeeper to
prevent irrelevant or unreliable testimony from being admitted.” Dobbs, 392
Wis. 2d 505, ¶43.

¶22 Our supreme court recently concluded that WIS. STAT. § 907.02, as
amended, continues to permit an expert witness to provide “exposition testimony

9
No. 2020AP983-CR

on general principles without explicitly applying those principles to, or even
having knowledge of, the specific facts of the case.” Dobbs, 392 Wis. 2d 505,
¶42. The court emphasized, however, that the admission of exposition testimony
is “not automatic,” and when deciding whether to admit such testimony, the circuit
court must consider, among other things, “whether the testimony will ‘fit’ the facts
of the case.” Id., ¶43. The court further explained that “[f]it ‘goes primarily to
relevance.’” Id., ¶44 (quoting Daubert, 509 U.S. at 591). As such, whether
exposition testimony “fits” a particular case “turns on whether it is ‘sufficiently
tied to the facts of the case’ such that ‘it will aid the jury in resolving a factual
dispute.’” Id. (quoting Daubert, 509 U.S. at 591). “Generalized expert testimony
that is factually disconnected from the case is inadmissible because it does not
assist the jury in rendering a verdict based on the material facts in issue.” Id.
(quoting Trout v. Milton S. Hershey Med. Ctr., 576 F. Supp. 2d 673, 677
(M.D. Pa. 2008)).

¶23 In Dobbs, the defendant sought to introduce an expert witness’s
general testimony regarding “the phenomenon of false confessions, as well as the
interrogation techniques and dispositional characteristics that make false
confessions more likely.” Id., ¶19. Dobbs argued this testimony would have
“assisted the jury in assessing the truthfulness of [his] confessions to police by
correcting a common misbelief that innocent people do not confess to crimes they
did not commit.” Id., ¶47.

¶24 Our supreme court rejected Dobbs’ argument, concluding the circuit
court did not erroneously exercise its discretion by excluding the expert’s
testimony. Id., ¶51. The court noted the circuit court had found that Dobbs was
not subjected to most of the coercive interrogation techniques that his expert
described. Id., ¶49. In addition, the court noted the record contained at least six

10
No. 2020AP983-CR

instances in which Dobbs had spontaneously admitted criminal activity absent any
coercive police tactics. Id. The court also stated the record showed that Dobbs
did not possess most of the characteristics that the expert testified may predispose
someone to confess falsely if coercive interrogation tactics are used. Id., ¶50. The
court stated that under these circumstances, the circuit court “could have
reasonably concluded that [the expert’s] exposition testimony regarding situational
factors that increase the likelihood of false confessions would not assist the trier of
fact to understand the evidence.” Id., ¶51. In other words, the circuit court could
have reasonably concluded that the expert’s proffered testimony did not
adequately “fit” the facts of the case.

¶25 Although it preceded Dobbs, our decision in State v. Schmidt, 2016
WI App 45, 370 Wis. 2d 139, 884 N.W.2d 510, is also instructive. Like this case,
Schmidt involved the admissibility of Thompson’s expert testimony regarding
factors that may affect the reliability of a child witness’s statements. Id., ¶¶55-60.
On appeal, we assumed, without deciding, that Thompson’s testimony was
admissible under WIS. STAT. § 907.02. Schmidt, 370 Wis. 2d 139, ¶75. We
concluded, however, that the circuit court had properly excluded the testimony on
relevance grounds. Id., ¶76. We explained that Thompson

could not opine whether any of the suggestive interview
techniques he discussed were actually used with [the child],
nor could he conclude the changes in [the child’s] story
over the course of several years were attributable to such
techniques. There were no visual or audio recordings of
the interviews for Thompson to review, no transcript of the
interviews, and Thompson never met or spoke with [the
child]. At best, Thompson could testify generally as to the
findings of the research regarding the suggestibility of
children who were often much younger than [the child].

Id., ¶75.

11
No. 2020AP983-CR

¶26 Stated differently, although Thompson proposed to testify in
Schmidt about how certain interview techniques and external factors can influence
a child’s memories, his testimony was not relevant because he “could only testify
about such matters at a high level of generality and could not tie these concepts in
any meaningful way to the particular circumstances surrounding [the child’s]
statements to police.” Id., ¶78. While Schmidt addressed the admissibility of
Thompson’s testimony through the lens of relevance, it essentially concluded that
the circuit court properly excluded his testimony because it did not “fit” the facts
of the case. The supreme court subsequently recognized in Dobbs that “fit”—a
requirement for the admission of expert exposition testimony under WIS. STAT.
§ 907.02—“goes primarily to relevance.” Dobbs, 392 Wis. 2d 505, ¶44 (quoting
Daubert, 509 U.S. at 591).

¶27 As noted above, Thompson’s report in this case identified six factors
that he asserted “affect[] the reliability of a child’s statements.” Thompson’s
report then discussed how those factors could have affected the reliability of Gail’s
report that Castillo sexually assaulted her. Thompson did not, however,
conclusively state that any of the factors discussed in his report were actually
present in Gail’s case. Moreover, at the Daubert hearing, Thompson expressly
conceded that those factors “could have affected the victim[,] but we don’t know if
they did.” Based on Thompson’s report and his testimony at the Daubert hearing,
the circuit court could reasonably determine that his proffered testimony should be
excluded because it did not adequately “fit” the facts of the case.

¶28 To explain further, the first factor identified in Thompson’s report
was repeated interviewing. In the report, Thompson noted that Gail had been
interviewed a minimum of three times before her forensic interview—at least once
by her mother, once by a physician’s assistant at the hospital, and once by a child

12
No. 2020AP983-CR

protective services worker.7 Thompson stated research has shown that “when
children are repeatedly interviewed about non-events … they report such events at
a relatively high rate and vigorously hold to those false memories despite evidence
to the contrary.” Thompson also acknowledged, however, that under some
circumstances “the use of repeated interviewing with children can be a useful tool
and can result in additional accurate information from the interviewee.”
Nevertheless, Thompson stated that advantage has not been demonstrated “in
situations where one or more of the repeated interviews utilized either a biased
interviewer or inappropriate interviewing techniques.”

¶29 Thompson therefore asserted that “[t]o the extent [Gail’s] mother
used suggestive or leading questions” when asking Gail about the assault, Gail’s
memories “may be tainted from that point forward.” Thompson likewise stated
that Gail’s recollections “may have been irreparably tainted from the outset” to the
extent she was asked “inappropriate or leading questions” by either her mother or
the physician’s assistant, and to the extent she and her sister discussed the
allegations amongst themselves. Thompson ultimately stated, however, that it was
“not possible for [him] to opine concerning the extent to which that actually
occurred, as [he had] no reports or recordings upon which to base such an
assessment.” In a similar vein, when discussing the effects of repeated
interviewing during his testimony at the Daubert hearing, Thompson conceded, “I
can’t tell you if that clearly affected this particular child or not.” When later asked

7
Thompson defined the term “interview” to include “conversations the child has with
formal interviewers, such as those conducted at child advocacy centers, as well as less structured
informal conversations that the child has with other persons,” including family members.

13
No. 2020AP983-CR

if he knew whether the repeated interviews in this case affected the reliability of
Gail’s report, Thompson responded, “I don’t know.”

¶30 The second factor identified in Thompson’s report was “external
influences.” Thompson explained in the report that external factors—such as
family pressure and overhearing others talk about an individual in a negative
manner—can have a significant impact on a child’s statements. He then noted that
Anne had told police she had “trust issues” with Castillo. He conceded, however,
that it was “unclear as to whether or not th[o]se concerns had been communicated
to [Gail].” Thompson also suggested that Gail may have obtained sexual
knowledge through interactions with a male child. In addition, he stated Gail’s
use of the term “rape” in her forensic interview suggested her report had been
affected by external influences.

¶31 During the Daubert hearing, however, when Thompson was asked
what effect external influences may have had on Gail’s memories, he
responded: “I can’t tell you exactly what the effects were because … that’s the
job for the trier of fact.” Thompson also confirmed during the Daubert hearing
that he did not have any information about whether Anne’s “trust issues”
regarding Castillo had been communicated to Gail. In addition, while Thompson
testified that external sources could have exposed Gail to sexualized behavior or
vocabulary, he did not “know exactly if that was the case.” He also conceded that
he did not know whether any sexualized interactions between Gail and a male
child had actually occurred. Ultimately, while Thompson testified there “may
have been some external influences” on Gail, he could not say “how that would
have formally affected her or specifically affected her memory and her reports.”

14
No. 2020AP983-CR

¶32 The third factor identified in Thompson’s report was inappropriate
interviewing techniques. Thompson explained in the report that “[t]he use of
leading and Yes/No questions, social reinforcement for specific statements, or
disapproval for specific statements all can have a significant impact on the
accuracy of a child’s reports.” Thompson then classified each “utterance[]” made
by the individual who conducted Gail’s forensic interview as either an invitation, a
directive, an option-posing utterance, or a suggestive utterance. Thompson
asserted that Gail’s interviewer had used directives and option-posing utterances
more frequently than what Thompson believed to be “best practices.” Thompson
opined that the use of those types of utterances is “likely to contribute to decreased
reliability of the child’s memory and statements” under circumstances where the
child has been “exposed to misinformation through repeated interviewing and
external influences.” Again, though, Thompson could not say whether Gail had
actually been exposed to such misinformation.

¶33 Moreover, while Thompson generally asserted that the interviewer’s
use of certain types of utterances did not comport with “best practices,” he did not
identify any specific utterances or questions that he believed were improper.
Thompson also conceded in his report that “[n]one of the queries posed by [the
interviewer] met the strict criteria for coding as suggestive.” At the Daubert
hearing, Thompson similarly testified that while there were “specific techniques
that this interviewer could have used to elicit more information,” the interviewer
“did not ask any suggestive questions.”

¶34 With respect to Thompson’s fourth factor—interviewer bias—
Thompson’s report explained that interviewers sometime engage in “confirmatory
bias,” which occurs when an interviewer “approaches an interview with a
preconceived notion of what may have occurred” and then “tends to pay attention

15
No. 2020AP983-CR

to or assign greater weight to information or statements that confirm the
interviewer’s preconceived ideas, while simultaneously paying less attention to or
giving less weight to statements that refute the interviewer’s preconceived notion.”
Thompson stated, however, that he “did not find evidence that suggested that
[Gail’s forensic] interviewer was biased.” In fact, he specifically noted that the
interviewer “appeared to make numerous attempts to test alternative hypotheses,”
which militated against a finding of bias.

¶35 As for Thompson’s fifth factor—therapy effects—his report
explained that “[m]any appropriate psychotherapeutic techniques that are used to
treat children that have experienced trauma, including various forms of abuse, can
affect the child’s recollections and reports.” Thompson acknowledged, however,
that he “did not find any specific evidence that [Gail was] involved in
psychotherapy.” At the Daubert hearing, Thompson noted that Gail had been
referred to therapy during the forensic interview, but he acknowledged that any
therapy she participated in thereafter would not have affected the reliability of
statements she made during the interview. Thompson testified he did not know
whether Gail had participated in therapy before the forensic interview.

¶36 Finally, the sixth factor identified in Thompson’s report was source
misattribution errors—that is, “situations in which an individual … identifies the
incorrect source of a memory.” Thompson’s report noted there had been
“allegations” that Gail “had engaged in sex play with … [a male child].”
Thompson theorized that such activity “may well [have served] as the basis for a
source misattribution error.” Thompson conceded, however, that it was unclear to
what extent Gail had “experienced this behavior [with the male child] in a
sexualized manner.”

16
No. 2020AP983-CR

¶37 Moreover, as noted above, during the Daubert hearing Thompson
admitted that he did not know whether any sexualized interactions between Gail
and a male child had actually occurred. Thompson also conceded that he could
not say whether Gail’s interactions with the male child affected the reliability of
her report that Castillo had assaulted her. Furthermore, the circuit court ruled that
any evidence regarding alleged “sexual play activity” between Gail and the male
child was inadmissible under the rape shield statute, and Castillo does not
challenge that ruling on appeal. As such, the jury would not have been permitted
to consider the sole basis Thompson identified in support of his assertion that a
source misattribution error may have occurred in this case.8

¶38 As the above discussion shows, while Thompson proposed to testify,
generally, about factors that can affect the reliability of a child witness’s
statements, there was little to no evidence showing that those factors were actually
present in this case. In addition, Thompson conceded during the Daubert hearing
that he did not know to what extent the factors he discussed had actually affected
the reliability of Gail’s statements.

8
Thompson did not identify any other sources of sexual knowledge that he believed
could have caused Gail to make a source misattribution error when accusing Castillo of sexual
assault. At trial, Castillo attempted to suggest that Gail may have gained sexual knowledge by
observing Anne and her boyfriend. The only evidence he presented in support of that theory,
however, was Lorena’s testimony that Gail had reported seeing Anne and her boyfriend under the
covers in a hotel room, and that the boyfriend was on top of Anne and they were making noises.
Even if the jury accepted Lorena’s testimony as true, Gail’s observation of that incident would
not have provided an explanation for her misattribution of the specific sexual acts that she
accused Castillo of committing.

Castillo also attempted to suggest at trial that Gail had gained sexual knowledge by
watching a television crime program with Lorena. However, no evidence was presented about
the specific contents of that program. As such, there would have been no basis for the jury to
conclude that it was the source of Gail’s report that Castillo had sexually assaulted her in the
specific ways she alleged.

17
No. 2020AP983-CR

¶39 The circuit court therefore reasoned that although Thompson was
“talking about external influences on a child’s memory … in this particular case
there are only hunches and guesses about what could have happened.” The court
further stated that because there was nothing “scientifically reliable to suggest that
[the factors Thompson cited] affected this child’s memory,” his proffered
testimony was “speculative” and did not “match up with the particular incident.”
In other words, the court concluded that Thompson’s proffered testimony should
be excluded under WIS. STAT. § 907.02 because it did not adequately “fit” the
facts of the case. Based on Thompson’s report and his testimony at the Daubert
hearing, we conclude the court did not erroneously exercise its discretion in that
regard.

¶40 In any event, even if expert testimony satisfies the standard for
admissibility in WIS. STAT. § 907.02 and is relevant under WIS. STAT. § 904.01, a
circuit court nevertheless has discretion to exclude the testimony under WIS. STAT.
§ 904.03. See Schmidt, 370 Wis. 2d 139, ¶86. Under that statute, relevant
evidence may be excluded “if its probative value is substantially outweighed by
the danger of unfair prejudice, confusion of the issues, or misleading the jury, or
by considerations of undue delay, waste of time, or needless presentation of
cumulative evidence.” Sec. 904.03.

¶41 In this case, the circuit court determined that Thompson’s proffered
testimony was “speculative”—i.e., lacking in probative value—and would be
“confusing to the jury.” Stated differently, the court determined that any probative
value Thompson’s testimony had would be substantially outweighed by the danger
of confusing the issues. The court did not erroneously exercise its discretion by so
concluding.

18
No. 2020AP983-CR

¶42 Again, our decision in Schmidt is instructive. There, we concluded
that “[w]hat minimal probative value Thompson’s testimony may have had
regarding [the child’s] credibility was easily outweighed by the very real potential
that Thompson’s testimony would mislead or confuse the jury by requiring them
to speculate about what had occurred during the police interviews and elsewhere.”
Schmidt, 370 Wis. 2d 139, ¶86. We reasoned that “if Thompson’s testimony were
admitted into evidence, it [was] entirely probable the jury would conclude, based
solely on the fact he was testifying, that suggestive interview techniques had been
used with [the child] despite the absence of any evidence to that effect.” Id.

¶43 The circuit court could reasonably conclude that the same is true
here. That is, without any admissible evidence showing that Gail was exposed to
any of the factors Thompson identified, the court could reasonably conclude that
Thompson’s testimony would have invited the jury to speculate that those factors
were present simply because he discussed them at a generalized level. As such,
even if the court erred by ruling that Thompson’s testimony was inadmissible
under WIS. STAT. § 907.02, the court did not erroneously exercise its discretion by
excluding his testimony under WIS. STAT. § 904.03. We therefore reject Castillo’s
argument that the court erred by excluding Thompson’s testimony.9

9
In the alternative, the State also argues that the circuit court properly concluded the
admission of Thompson’s testimony would violate State v. Haseltine, 120 Wis. 2d 92, 352
N.W.2d 673 (Ct. App. 1984), by improperly offering an opinion on Gail’s credibility. Because
we conclude the court did not erroneously exercise its discretion by excluding Thompson’s
testimony on two other grounds, we need not address this alternative argument. See Turner, 268
Wis. 2d 628, ¶1 n.1.

19
No. 2020AP983-CR

II. Mistrial

¶44 Castillo next argues that the circuit court erred by denying his
motions for a mistrial. When confronted with a request for a mistrial, the circuit
court must determine, “in light of the whole proceeding, whether the basis for the
mistrial request is sufficiently prejudicial to warrant a new trial.” Bunch, 191
Wis. 2d at 506. In other words, the court must determine whether the defendant
can receive a fair trial, in light of all the facts and circumstances. State v. Ford,
2007 WI 138, ¶29, 306 Wis. 2d 1, 742 N.W.2d 61. The decision to grant or deny a
mistrial lies within the circuit court’s sound discretion, and we will reverse only if
there is a clear showing that the court erroneously exercised its discretion. Bunch,
191 Wis. 2d at 506. Moreover, in a case like this one where the defendant’s
request for a mistrial was not based on any laxness or overreaching by the
prosecution, we must give great deference to the circuit court’s ruling. Id. at 507.

¶45 As discussed above, Castillo first moved for a mistrial after the
prosecutor asked Gail if there was “anything else that you remember about what
happened that you think we should know,” and Gail responded, “He did it to three
other little girls.” Castillo again moved for a mistrial after his sister Lorena
testified that she and Castillo had discussed “him getting a job if he gets out.”
Castillo argues—and the State does not appear to dispute—that both of these
statements violated the circuit court’s pretrial evidentiary rulings. Castillo
correctly notes that “[a]n important element of a fair trial is that a jury consider
only relevant and competent evidence bearing on the issue of guilt or innocence.”
See Bruton v. United States, 391 U.S. 123, 131 n.6 (1968).

¶46 Castillo acknowledges that the circuit court struck the testimony in
question and instructed the jury to disregard it, and he concedes that we generally

20
No. 2020AP983-CR

presume jurors follow the court’s instructions. See Truax, 151 Wis. 2d at 362. He
asserts, however, that “some statements are so prejudicial, they cannot be
remedied by a curative instruction and cannot be unheard by the jury.” Castillo
contends the statements at issue in this case fall into that category. He argues that
“simply asking the jury to disregard” a scared, eight-year-old girl’s assertion that
the person accused of sexually assaulting her had also assaulted three other little
girls “would not be enough to cure the defect,” given the highly prejudicial nature
of that comment. Castillo also argues that statements regarding a defendant’s
incarceration are “inherently prejudicial” and are comparable to allowing the jury
to see the defendant in “jail garb.” See Estelle v. Williams, 425 U.S. 501, 504-05
(1976) (explaining that a defendant’s appearance at trial in jail attire “may affect a
juror’s judgment” and presents “an unacceptable risk ... of impermissible factors
coming into play”). Castillo thus asserts that although the jury did not see him in
jail garb, “he nonetheless was prejudiced by the reference to previously being
incarcerated.”

¶47 Moreover, Castillo argues that the two objectionable statements,
taken together, had the “combined effect” of conveying to the jury that he was “a
bad and dangerous individual.” He asserts that “without anyone in the courtroom
in the position to clarify the facts of the matter,” the jury was “left to speculate
whether the two struck statements were related.” He contends, “Taken together,
jurors could assume that Mr. Castillo had been incarcerated previously due to prior
sexual assault convictions and that he is a serial child abuser.” Castillo further
asserts that the combined effect of the two impermissible statements was
“particularly problematic” because the case was essentially a “credibility battle”
between Gail and Castillo, and it is likely the statements negatively affected the
jury’s assessment of Castillo’s credibility.

21
No. 2020AP983-CR

¶48 In response, the State emphasizes our standard of review, and the
deference we must give the circuit court’s decision to deny a mistrial when the
mistrial request was not occasioned by any prosecutorial laxness or overreaching.
The State asserts the court’s oral rulings show that it properly exercised its
discretion by denying both of Castillo’s mistrial motions. The court determined
that the prosecution had not deliberately elicited either of the improper comments.
With respect to Gail’s comment, the court stated the jury would “understand[] that
a child might blurt something out.” As for Lorena’s comment, the court
emphasized that Lorena was a defense witness and was hostile to the prosecution.
In both instances, the court determined the jury was capable of following the
court’s instruction to disregard the improper remark.

¶49 The State also emphasizes the “legal presumption that the jury
adhered to the court’s limiting instructions.” The State notes that the circuit court
immediately struck both of the improper comments and instructed the jury to
disregard them. The State also observes that after the close of evidence, the court
instructed the jury that it should disregard all stricken testimony, that it should
consider evidence of witnesses’ prior crimes only in order to evaluate their
truthfulness, and that Castillo’s prior juvenile delinquency adjudications were
relevant only as to his credibility and were not “proof of guilt of the offense now
charged.”

¶50 The State argues Castillo has not overcome the presumption that the
jury followed these instructions. With respect to Gail’s statement, the State asserts
Castillo fails to acknowledge three “mitigating factors” that lessened the
statement’s prejudicial effect: (1) the jury’s understanding that a scared child
“might blurt something out” while testifying; (2) the fact that no other evidence or
mention of “other little girls” was introduced at trial; and (3) the circuit court’s

22
No. 2020AP983-CR

instruction to the jury that it could not infer guilt based on evidence of other
crimes.

¶51 As for Lorena’s statement about Castillo getting a job “if he gets
out,” the State contends the jury would not necessarily have interpreted that
remark to mean that Castillo was incarcerated. In any event, the State asserts that
even if the jury understood Lorena’s comment in that way, there was “no
suggestion that he was incarcerated pending trial in this case, as opposed to one of
his other 15 delinquency adjudications that the jury was informed about.” The
State further contends that the prejudicial effect of a jury inferring that a defendant
was incarcerated shortly before trial is not the same as the “visual impact of seeing
the defendant in clothing or circumstances that implies guilt.”

¶52 Finally, the State argues that the prejudicial effect of Lorena’s
statement was diminished by the fact that the jury could have inferred Castillo was
incarcerated based on other evidence admitted at trial—namely, Castillo’s
testimony that he had fifteen prior juvenile delinquency adjudications; Lorena’s
testimony the she knew her phone calls with Castillo were recorded; and Lorena’s
statement that “when it’s our visits, we talk about what we want out of this.” The
State contends the jury “could easily have inferred that Castillo was incarcerated
before trial based on these unobjected-to statements alone.” The State therefore
asserts that “the chance of any additional prejudice by hearing an offhand
reference to him ‘get[ting] out’ is nonexistent.”

¶53 After considering both parties’ arguments, we conclude the circuit
court erroneously exercised its discretion by denying Castillo’s motions for a
mistrial. Specifically, we agree with Castillo that, taken together, Gail’s statement
that Castillo “did it to three other little girls” and Lorena’s statement about Castillo

23
No. 2020AP983-CR

“get[ting] out” were sufficiently prejudicial as to prevent Castillo from receiving a
fair trial. Indeed, we are particularly troubled by Gail’s statement, given its
relation to the nature of the allegations in this case.

¶54 Evidence of other crimes is prejudicial, in part, because it results in
an “overstrong tendency to believe the defendant guilty of the charge merely
because he [or she] is a person likely to do such acts.” Whitty v. State, 34 Wis. 2d
278, 292, 149 N.W.2d 557 (1967). In addition, evidence of other crimes gives rise
to the danger that the jury will convict “not because [the defendant] is believed
guilty of the present charge but because he [or she] has escaped punishment from
other offenses.” Id. Here, the jury heard a scared,10 eight-year-old girl testify not
only that Castillo had sexually assaulted her, but also that he had done the same to
three other “little girls.” Gail’s comment to that effect created a substantial danger
that the jurors decided to convict Castillo of the assault alleged in this case
because they believed he had assaulted other children in the past.

¶55 As noted above, there was no physical evidence that Castillo
assaulted Gail, nor was there any witness who directly corroborated either Gail’s
or Castillo’s version of events. As such, the crucial issue at trial was whether the
jury found Gail or Castillo to be more credible. In the context of this credibility
contest, Gail’s statement about Castillo sexually assaulting other little girls was

10
Both Castillo and the State characterize Gail as having been scared during her trial
testimony. The record supports that characterization. At one point during her testimony, Gail
indicated that she was scared to answer the prosecutor’s question about how she knew that
Castillo was trying to put his private part in her bottom. When the prosecutor asked why Gail
was scared to answer that question, Gail responded, “There’s so much people.” When the
prosecutor later asked Gail what Castillo had told her to do to his private part, Gail stated she was
“embarrassed” to say the word. Later, Gail testified she did not like talking about what happened
to her in the bedroom and she was scared to tell people about it.

24
No. 2020AP983-CR

particularly problematic. Absent that statement, the jury may have had doubts
about whether to believe Gail’s or Castillo’s testimony. There is simply too great
a danger, however, that having heard Gail’s accusation that Castillo had sexually
assaulted other little girls, the jurors decided to credit Gail’s testimony about the
alleged assault simply because they believed Castillo had committed similar
crimes before. Although the State correctly notes that no other mention or
evidence of “other little girls” was introduced at trial, that fact does not erase our
concern that Gail’s statement prevented Castillo from receiving a fair trial, given
the highly prejudicial nature of the statement, the manner in which it was
delivered, and its relation to the charged conduct.

¶56 The circuit court noted that Gail was only eight years old at the time
of trial. Given her age, the court reasoned the jury was “capable of understanding
that a child might blurt something out” and would therefore be capable of
“disregarding the comment.” The State advances the same argument on appeal.
We do not find this analysis convincing. While the jury might understand the
possibility that a child witness would “blurt something out,” that does not mean
the jurors would necessarily disregard or disbelieve the child’s statement. Indeed,
the fact that Gail was a scared, eight-year-old child who had “blurt[ed] … out” an
accusation about Castillo sexually assaulting three other girls would serve to
enhance, not diminish, the credibility of that accusation. Gail’s young age and the
apparent spontaneity of her statement would militate against a conclusion that she
was making a calculated attempt to shock or improperly influence the jury by
fabricating an allegation about Castillo sexually assaulting other girls.

¶57 In any event, we agree with Castillo that the prejudicial effect of
Gail’s comment was compounded by Lorena’s subsequent statement about
Castillo “get[ting] out.” Taken together, the two statements could have led a

25
No. 2020AP983-CR

reasonable jury to conclude that Castillo had been incarcerated prior to trial
because he had sexually assaulted other little girls. Stated differently, Lorena’s
statement suggesting that Castillo was incarcerated enhanced the credibility of
Gail’s statement about his sexually assaulting other girls by giving rise to a
reasonable inference that Castillo was incarcerated as a result of those acts. The
jury’s awareness of Castillo’s fifteen prior juvenile delinquency adjudications
would have further supported that inference. Thus, Lorena’s statement contributed
to the likelihood that the jury convicted Castillo not because it was convinced of
his guilt based on the admissible evidence presented at trial, but because it
believed he was a serial child abuser and therefore must have committed the
assault alleged in this case.

¶58 On these facts, we conclude Castillo has overcome the presumption
that the jury followed the circuit court’s instructions to disregard Gail’s and
Lorena’s statements. We agree with Castillo that the prejudicial effect of those
statements, taken together, was so great that the court’s instructions were “not
sufficient to remedy the error[,] and the evidentiary bell could not be unrung.”
“[I]f you throw a skunk into the jury box, you can’t instruct the jury not to smell
it.” Dunn v. United States, 307 F.2d 883, 886 (5th Cir. 1962). Here, the
proverbial odor of the improper statements was so strong that we cannot conclude
the court’s cautionary instructions were sufficient to prevent the jury from
considering those statements and allowing them to affect its deliberations.

¶59 The State asserts the circuit court found, based on its “observation of
the jury and the effect of the comments,” that the jury “would follow the court’s
instructions and disregard the statements.” The State then suggests that we must
defer to the court’s determination in that regard. We acknowledge the deferential
standard of review that we must apply in this appeal. See Bunch, 191 Wis. 2d at

26
No. 2020AP983-CR

506-07. We also acknowledge that a circuit court’s factual findings are not set
aside on appeal unless they are clearly erroneous. See Phelps v. Physicians Ins.
Co. of Wis., 2009 WI 74, ¶34, 319 Wis. 2d 1, 768 N.W.2d 615. Here, however,
the court made no factual findings regarding the jury’s demeanor or the effect of
the objectionable statements on the jurors. The court simply stated that it believed
the jury was capable of disregarding the statements. We disagree, for the reasons
explained above. In short, the statements simply posed too great a risk that the
jury would convict Castillo based on a belief that he was a serial child abuser,
rather than on the admissible evidence introduced at trial.

¶60 We therefore conclude that the circuit court erroneously exercised its
discretion by denying Castillo’s requests for a mistrial. Given the highly
prejudicial nature of Lorena’s, and especially Gail’s, challenged statements, and in
light of all the facts and circumstances, the only reasonable conclusion is that the
introduction of the statements prevented Castillo from receiving a fair trial. See
Ford, 306 Wis. 2d 1, ¶29. Accordingly, we reverse Castillo’s judgment of
conviction and remand for a new trial.

By the Court.—Judgment reversed and cause remanded for further
proceedings.

Not recommended for publication in the official reports.

27